How Epistemic Modifiers Emerge 9783110411027, 9783110400564

Citation preview

Jakob Maché How Epistemic Modifiers Emerge

Trends in Linguistics Studies and Monographs

Editors Chiara Gianollo Daniël Van Olmen Editorial Board Walter Bisang Tine Breban Volker Gast Hans Henrich Hock Karen Lahousse Natalia Levshina Caterina Mauri Heiko Narrog Salvador Pons Niina Ning Zhang Amir Zeldes Editor responsible for this volume Volker Gast

Volume 292

Jakob Maché

How Epistemic Modifiers Emerge

ISBN 978-3-11-040056-4 e-ISBN (PDF) 978-3-11-041102-7 e-ISBN (EPUB) 978-3-11-041112-6 ISSN 1861-4302 Library of Congress Control Number: 2018963772 Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available on the Internet at http://dnb.dnb.de. © 2019 Walter de Gruyter GmbH, Berlin/Boston Printing and binding: CPI books GmbH, Leck www.degruyter.com

| For Susanne Benesch ⁕ 22th July 1898, Vilnius † Spring 1940, Ravensbrück

But your seed has prospered . . .

Acknowledgements The present book is a revised, updated and refined version of my dissertation, which was defended at the Freie Universität Berlin on 28th June 2013. A lot of people contributed in direct an indirect ways. First of all, I would like to thank the Austrian and the German tax payers, without whose help it would have been impossible to pursue an extensive investigation such as the one presented here. The earlier parts of this work were funded by the Österreichische Akademie der Wissenschaften (Austrian Academy of Science), which supported me with a DOC grant from May 2005 to April 2007. The rest is the result of my research carried out at the Freie Universität Berlin from October 2007 to the Winter 2013. There are a couple of people who provided me with crucial support and to whom I am deeply indebted. First of all, I would like to thank Werner Abraham, who opened many doors to me that would have been remained closed otherwise, and Stefan Müller for being so trusting when he employed me as his research and teaching assistant. Further important facilitators are Ulrich Dressler and Richard Schrodt. Aside from that, there are a lot of other people who contributed to the present study by giving fruitful hints or criticism. Among the most important partners are my colleagues Tanyushichka Ackermann, Katrin Axel, Felix Bildhauer, Rui Chaves, Philippa Cook, Katarina Colomo, Hamida Demirdache, Kristin M. Eide, Serge Doitchinov, Markus Egg, Veronika Ehrich, Peter Eisenberg, Ophira Gamliel, Volker Gast, Remus Gergel, Jane Grimshaw, Nele Hartung, Roland Hinterhölzl, Anke Holler, Katalin Horváth, Łukasz Jędrzejowski, Olga Kellert, Ekkehard König, Zhanna Lipenkova, Claudia Maienborn, Amalia Mendes, Mingya Liu, Svetlana Petrova, José Pinto de Lima, Kilu von Prince, Uli Reich, Marga Reis, Eva-Maria Remberger, Roland Schäfer, Oliver Schallert, Mathias Schenner, Cosima Scholz, Horst Simon, Augustin Speyer, Hubert Truckenbrodt, Evdokia “the one and only” Valcheva, Heinz Vater, Rita Veloso, Hedde Zeijlstra, Chris Zimmer. I am particularly indebted to Fé.mi Adéwo.lé from the O.báfé.mi Awóló.wò. University Ilé-Ifè. for sharing with me his unique expertise on modal verbs in Yorùbá. Furthermore, I am grateful for the comments and hints given by Anastasios Tsangalidis, Hardarik Blühdorn, Jonathan Ginzburg, Joachim Jacobs and Horst Lohenstein and possibly many more. Finally, I would like thank Konstantin Krasuchin for the fruitful discussions. Moreover, I would like to thank those native speakers of English who helped me with the multiple proof-readings of this 530 pages long monster. In particular, Erica Haas, Philippa Cook, Kevin Ryan and Natalie Adams. Yet, above all, I am grateful to Volker Gast, the editor of this series, who over many months and years https://doi.org/10.1515/9783110411027-201

VI | Acknowledgements

invested so much time to personally proof-read the whole manuscript for another time. In addition, Till Kulawik deserves special thanks for helping me with analysing the data from Latin. Apart from that, there are a couple of people who helped me to bring various research visits to fruition. At this point I would like to thank Elisabeth Leiss and Christiane Wanzeck for supporting me on my visit at the Ludwig-MaximiliansUniversität München in September 2005, Marga Reis for hosting me at the Sonderforschungsbereich 441 Linguistische Datenstrukturen at the Universität Tübingen in the years 2005/2006, Amalia Mendes and Ana Maria Martins for my stay at the Centro de Linguística da Universidade de Lisboa in 2006, and Hamida Demirdache for my stay at the Laboratoire Linguistique de Nantes at the Université de Nantes in May 2007. I am particularly indebted to Viola Auermann, who supported me in many difficult situations and who enabled me to deliver this thesis on time. Finally, I owe Nita Little, Nancy Stark Smith and Steve Paxton a great debt of gratitude for their revolutionary inventions. And last but not least, I am grateful to the Estádio do Luz for providing me with divine insights from 2007. And to my bicycle, which carried me through Berlin every day, altogether more than 40.000 kilometres.

Contents Acknowledgements | V List of Tables | XI Abbreviations | XIII 1 1.1 1.2 1.3 1.4

Introduction | 1 Aims | 1 Method | 1 Theoretical considerations | 3 Results in a nutshell | 4

2 Modal verbs: A class struggle | 8 2.1 Traditional criteria | 9 2.1.1 Morphological criteria | 9 2.1.2 Syntactic criteria | 20 2.1.3 Semantic criteria | 25 2.1.4 Conclusions | 33 2.2 Case studies | 37 2.2.1 können | 38 2.2.2 müssen | 75 2.2.3 wollen | 95 2.2.4 dürfen | 113 2.2.5 dürfte | 122 2.2.6 sollen | 129 2.2.7 mögen | 147 2.2.8 möchten | 165 2.2.9 brauchen | 176 2.2.10 werden | 220 2.2.11 scheinen and dünken | 229 2.2.12 drohen, versprechen and verheißen | 238 2.2.13 Summary | 250 2.3 The origin of the term Modalverb | 256 2.3.1 Early grammars: A morphological classification | 257 2.3.2 Karl Ferdinand Becker (1836): From a morphological classification to a grammatical one | 260 2.3.3 Blatz (1896) | 264

VIII | Contents

2.3.4 2.3.5 2.3.6 2.3.7 2.3.8

George O. Curme | 265 Braune, Paul and Behaghel: The decline of the semantic motivation | 266 Gunnar Bech (1949) | 268 The Duden grammar | 270 Summary | 273

3 The nature of epistemic modality | 280 3.1 Types of modification | 280 3.2 Circumstantial modal verbs as event modifiers | 286 3.3 The case of can | 287 3.4 Ambiguities across categories | 289 3.4.1 Conditionals | 289 3.4.2 Causal clauses | 294 3.4.3 Causal wo-clauses | 298 3.4.4 Corrective uses of obwohl, wobei and wiewohl | 299 3.4.5 Contrastive während-clauses | 301 3.4.6 so lange-clauses | 302 3.4.7 bevor-clauses | 303 3.4.8 Final damit-clauses | 304 3.4.9 Addressee-oriented free relative clauses | 306 3.4.10 Adverbial infinitives | 306 3.4.11 VP adverbs | 309 3.4.12 Locative modifiers | 313 3.4.13 noch einmal | 314 3.4.14 Past operator | 317 3.4.15 Meta-communicative why | 318 3.4.16 Declarative questions | 319 3.4.17 Summary | 320 3.5 Ambiguities across languages | 323 3.6 Critical data | 329 3.6.1 Quantificational modal verbs | 329 3.6.2 Intensional subjects | 330 3.6.3 Veronika Ehrich’s counter example | 331 3.6.4 Summary | 333 4 4.1 4.2 4.3

Twenty-one commandments for epistemic modality | 334 No infinitives | 338 No past participle | 342 No past tense | 348

Contents |

4.4 4.5 4.6 4.7 4.8 4.9 4.10 4.11 4.12 4.13 4.14 4.15 4.16 4.17 4.18 4.19 4.20 4.21 4.22 4.22.1 4.22.2 4.22.3 4.22.4 4.22.5 4.22.6 4.22.7 4.23 5 5.1 5.1.1 5.1.2 5.1.3 5.1.4 5.1.5

Excluded from the scope of a counterfactual operator | 355 Excluded from nominalisations | 357 No verbless directional phrase complements | 360 No VP-anaphora | 361 No separation in wh-clefts | 364 May not bear sentence accent | 368 Excluded from the scope of negation | 372 Excluded from polarity questions | 380 Excluded from wh-questions | 389 Excluded from imperatives | 395 Excluded from optatives | 396 Excluded from complement clauses | 398 Excluded from event-related causal clauses | 415 Excluded from the antecedent of an event-related conditional | 425 Excluded from temporal clauses | 433 Excluded from restrictive relative clauses | 441 Excluded from the scope of a quantifier | 444 No assent/dissent | 459 Objective and subjective epistemic modality: A reassessment | 460 Lyon’s original motivation | 461 Further advancements in the study of ‘objective’ epistemic modality | 466 The role of public evidence | 469 Objective epistemic modal verbs do not constitute a consistent class | 470 ‘Objective’ modal verbs that are practical possibility or quantificational modal verbs | 475 ‘Objective’ epistemic modal verbs that are ‘subjective’ | 476 Conclusions | 478 Summary | 480 Reportative and evidential modal operators | 484 Reportative wollen and sollen | 484 Infinitives | 485 Past participles | 488 Nominalisations | 489 Optatives | 491 Past tense | 491

IX

X | Contents

5.1.6 5.1.7 5.1.8 5.2 5.2.1 5.2.2 5.2.3 5.2.4

Questions | 495 Event related conditional clauses | 499 Summary | 500 So-called ‘evidentials’ drohen, versprechen and scheinen | 502 Past participles | 502 Event related conditional clauses | 509 Past tense | 512 Summary | 515

6 6.1 6.1.1 6.1.2 6.1.3 6.1.4 6.1.5 6.2 6.2.1 6.2.2 6.2.3 6.3 6.3.1 6.3.2 6.4 6.5

Anchoring the deictic centre | 517 The speaker, the addressee and arguments | 517 Declarative speech acts | 518 Interrogative speech acts | 518 Complement clauses | 519 Reportative modal verbs | 520 Summary | 521 The deictic centre | 521 The modal source | 522 The Condition on Deictic Centres | 523 A subtle refinement | 529 A unified analysis for epistemic and reportative modality | 532 Hierarchy of Salience | 532 Operators which impose selectional restrictions | 533 Alternative analyses | 536 Summary | 539

7 7.1 7.2 7.3

On black magic: A diachronic explanation | 540 Epistemic modal verbs in Early New High German | 541 The Rule of Accommodation as a driving force of language change | 548 Summary | 549

8

Summary | 551

Primary sources | 555 References | 559 Author index | 577 Subject index | 583

List of Tables

Tab. 2.5 Tab. 2.6 Tab. 2.7

Preterite present origin of wissen — the Old High German paradigm | 10 Modal verbs and preterite present morphology | 12 Most frequent verbs in spoken German, according to Ruoff (1981), based on a corpus study comprising 500,000 word form tokens | 36 Frequency of auxiliaries among all word form tokens, according to Kaeding (1897), based on a corpus study comprising 10,910,777 word form tokens | 37 Complement types of potential modal verbs | 254 Different extensions of the term Modalverb accros the centuries | 274 Features relevant for a definition of the class Modalverb | 279

Tab. 4.1 Tab. 4.2

Epistemic modal verbs in non-canonical enviroments | 474 Anchoring of epistemic modal operators in embedded contexts | 481

Tab. 7.1

Circumstantial modal verbs with stative complements – in Schmid’s Neuwe Welt (1567) | 545

Tab. 2.1 Tab. 2.2 Tab. 2.3 Tab. 2.4

https://doi.org/10.1515/9783110411027-203

Abbreviations The abbreviations employed in the glossed examples correspond to those found in the Leipzig Glossing Rules.¹ Additional tags used in the glossed examples are listed below: caus comp cor dim ger intn part ppp(ge) ppp(ipp) sup vpan

causative comparative correlate diminutive gerundive intensifier particle past participle with ge-prefix past participle with infinitivus pro participio-morphology superlative VP anaphora

1 The Leipzig Glossing Rules can be found at the following web site (last access 18th August, 2018): http://www.eva.mpg.de/lingua/resources/glossing-rules.php https://doi.org/10.1515/9783110411027-204

1 Introduction 1.1 Aims The primary aim of this investigation is to provide an analysis of the development of epistemic modal verbs in German. As Abraham (2002: 24) and Mortelmans, Boye and Auwera (2009) have illustrated, German has a particularly rich system of modal verbs, with a much richer morphology than is found in English. As has been shown by numerous investigations, such as Traugott (1989: 35), Öhlschläger (1989: 133), Sweetser (1990), Bybee, Perkins and Pagliuca (1994: 195), Fritz (1997), Diewald (1999) and Axel (2001: 45), the epistemic modal verbs of German developed diachronically from circumstantial modal verbs. However, the details of this development are not entirely clear, and figuring them out is no trivial matter. In order to tackle this issue, a couple of related questions have to be addressed beforehand. The first question that arises concerns the nature of the so-called modal verbs: How can these verbs be characterised? Is there a way to delimit them from other verbs? As will be shown, the availability of an epistemic interpretation plays an essential role for the classification of modal verbs. This raises another issue that has to be addressed regarding the nature of epistemic modality: How does epistemic modality differ from circumstantial modality? Are there environments in which epistemic modal verbs are used while their circumstantial counterparts are ungrammatical? Are there configurations in which epistemic modal verbs cannot occur while circumstantial modal verbs can? How can the distributional differences between epistemic modal verbs and circumstantial modal verbs be accounted for (if there are any)? Once it is clear what the precise nature of epistemic modality is, it will be possible to address the major question: What circumstances triggered the development of epistemic modal verbs? Summing up, the present investigation intends to provide answers to the following central questions: 1. What is the nature of modal verbs? 2. What is the nature of epistemic modifiers? 3. What triggered the development of epistemic modal verbs?

1.2 Method The use of modal verbs is one of the most extensively investigated phenomena in German. Accordingly, the number of descriptions and analyses is vast. However, https://doi.org/10.1515/9783110411027-001

2 | 1 Introduction

most of these accounts date from a period when no comprehensive electronic corpora were available. As a consequence, the large majority of previous investigations are based on very small sets of empirical data. Frequently, the data was obtained through introspection, a method which is not considered reliable anymore in contemporary linguistics. Moreover, a lot of analyses are based on scarce authentic data. Very often, the decisive examples on which the relevant theories are based involve configurations which are somewhere in the twilight zone of grammaticality and, thus, difficult to evaluate. In approaches that are supported by introspective data, such configurations are often judged as ungrammatical in the case of doubt. Yet, it often turns out that such allegedly ungrammatical configurations indeed exist if sufficiently large collections of texts are considered. Summing up, there are countless analyses of modal verbs in German that are most often based on introspective rather than on authentic data. In most cases, the grammaticality judgements of the decisive examples are fairly contested. Accordingly, their status as evidence in support of theoretical analyses is not always obvious. It is in this spirit that the present investigation provides selected data taken from the German Reference Corpus (Deutsches Referenzkorpus – DeReKo) composed and hosted by the Institut für Deutsche Sprache (IDS). At the time of investigation, i.e. during the years 2010–2012, it encompassed about 2 billion of word form tokens. A major contribution of this book is a systematic and thorough description of all potential modal verbs in German, which is well grounded on authentic data. Furthermore, all of the case studies will deliver a critical summary of the previous discussion on the respective verbs and refer to the most important approaches that have been published in the last century. In this way, the present investigation aims to provide an overview of previous accounts as well as a sound empirical foundation for future studies, rather than contributing another analysis that lacks the support of authentic data. This strategy may not appear very prestigious; however, the less spectacular way is often the more successful one. As has been noted above, this study has a diachronic focus. Diachronic studies involve a comparison of at least two historically distinct stages of a language. In the present study, the first stage concerns the time before the so-called modal verbs had acquired their epistemic interpretations, and the second stage covers the period after the modal verbs had developed their epistemic interpretations. Assuming that the latter stage corresponds to the situation in present-day German, one stage can already be clearly identified. As Fritz (1991: 29), Fritz (1997: 9, 95) and Müller (2001) illustrate, the number of uses of epistemic modal verbs attested in documents significantly increases for texts from the 16th century. Correspondingly, it appears to be likely that the 16th century plays an important role in the development of epistemic modal verbs.

1.3 Theoretical considerations |

3

In order to arrive at reliable results, any diachronic investigation needs to collect as much information about the respective stages to be contrasted as possible. In the present case, there is a bias between the two stages to be investigated. For present-day German, native competence can still be accessed. In contrast, no speaker with native competence for Early New High German or Middle High German can be found anymore. For this reason, it appears to be most natural to gain as much information as possible about the language stage that can still be accessed. In order to entirely understand the nature of a diachronic change, it is necessary to know at least one stage in every detail. Yet, there was no satisfactory description of the potential modal verbs in German that was based on authentic data at the time when this investigation was started. It appeared necessary to establish such a description before a reliable diachronic comparison could be undertaken. A similar approach is taken by Krämer (2005: 1). Moreover, the diachronic developments of a linguistic item are often reflected in the synchronic situation of a language. Ambiguous words often involve two or more variants that have arisen in different periods and that co-exist in the synchronic state. As Diewald (1999: 4) pointed out, that is exactly what applies to the ambiguity of modal verbs in German. Most modal verbs are ambiguous between transitive, circumstantial and epistemic uses and all of them have developed in different periods: The transitive was the source for the circumstantial patterns and the circumstantial patterns in turn constituted the basis for the epistemic uses. As the diachronic development of epistemic modal verbs is partially reflected in the synchronic state, we can learn a lot by investigating data from present day German. Accordingly, the investigation outlined here focusses on synchronic data and takes into account data from earlier periods whenever this provides relevant insights. In addition to that, grammars from the 17th and 18th centuries have been investigated to deepen the knowledge of the nature of the potential modal verbs in earlier stages. Finally, it will be argued that the German term Modalverb is misleading and inconsistent. In order to discover the source of these inconsistencies, the history of the term will be meticulously investigated across grammars from the 17th century up until the most recent decades.

1.3 Theoretical considerations Any theory is a model of reality. A theoretical model is the more successful the more it resembles reality. In linguistic theories, a particular language L is generally considered as the set of all grammatical sentences that can theoretically be uttered in this language. A perfect theoretical description of that language L yields

4 | 1 Introduction

the set that contains all those sentences that are judged as grammatical by the speakers of that language, and none that are judged as ungrammatical. As easy as it sounds, the way to the perfect description quickly turns out to be barred by uncountable dangerous obstacles paved with insidious traps – and sometimes, fallacious notions will mislead the eager scholar. In accordance with these prerequisites, the present investigation attempts to formulate a theory that captures as many uses of the so-called modal verbs in German as possible. It aims to cover of all the uses of the different (so-called) modal verbs that have hitherto been discussed in the literature and that can be found in corpora. In some of the more recent accounts, e.g. in usage-based theories, the grammaticality of a linguistic structure is occasionally related to its frequency in corpora. As these theories sometimes conclude, patterns that do not occur frequently are grammatical to a lesser extent, or not grammatical at all, and as a consequence, such uses can be neglected. Yet, frequency is not everything. Among rare linguistic structures, there are some that are regarded as deviant by the majority of the speakers of that language, but there are also instances that are considered as fully grammatical. In the latter case, the low frequency of a structure must obviously be due to some reason other than a failure to produce utterances of the relevant types.

1.4 Results in a nutshell Based on the three questions formulated in Section 1.1, the investigation of the corpus data has yielded the following results. As will be shown in Chapter 2, the term modal verb as used traditionally is not consistent. First of all, no characteristic could be found that separates the six traditional modal verbs, können ‘can’, müssen ‘must’, wollen ‘want’, dürfen ‘be allowed to’, sollen ‘shall’ and mögen ‘may’ from the remaining verbs in German. Furthermore, each of these verbs has turned out to behave in a very idiosyncratic manner. This illustrates that the traditional six modal verbs do not form a natural class, even if they exhibit some degree of undeniable kinship. The term modal verb as used traditionally suggests that its class members are characterised by two properties: They exhibit a morphological anomaly, and they fulfil the same function in grammar. Yet, the set of verbs with morphological anomalies and the set of verbs that denote a necessity or possibility are not co-extensive. Accordingly, the most efficient solution is to refrain from using the traditional term modal verb, and to restrict the focus to the epistemic patterns. In doing so, the extension of the class becomes larger, as it also includes verbs that

1.4 Results in a nutshell |

5

are not traditionally considered as modal verbs, e.g. werden ‘will, fut.aux’ and brauchen ‘need’. A similar approach has been taken by Reis (2001: 308, 2005). As it turns out, the epistemic uses of the verbs considered here constitute a natural class of verbs in German in formal and functional respects: They select bare infinitive complements and they can encode epistemic modality. It is reasonable to assume that these two properties are closely related to each other. As can be seen, the ‘ideal’ epistemic modal verb in German selects bare infinitive complements, and any verb that is about to acquire an epistemic interpretation has to lose its infinitive particle zu first. If the availability of an epistemic interpretation becomes the decisive property, the extension of the class has to be adapted. The class of epistemic modal verbs thus encompasses the following items: kann, könnte, muss, müsste, sollte, dürfte, mag, braucht nicht, and wird ‘will’. Due to the high number of idiosyncrasies that these verbs exhibit in their non-epistemic patterns, an analysis that is capable of capturing all these fine-grained differences is needed. It requires a lexicon that can differentiate between all the syntactic differences that the different potential modal verbs exhibit. For such an endeavour, a lexicalist account such as Head-Driven Phrase Structure Grammar seems to be the most promising one. As the availability of epistemic modality plays a crucial role in the classification of the verbs investigated here, it becomes necessary to understand its precise nature. As will be shown, there are characteristic contrasts between epistemic modal verbs and their circumstantial counterparts. Chapter 3 focusses on the environments in which only epistemic modal verbs are grammatical whereas their circumstantial cognates are ruled out. It will be pointed out that circumstantial modal verbs are event modifiers and, as a consequence, they are restricted to the selection of predicates that can be interpreted as events. By contrast, epistemic modal verbs can also embed predications about an identified subject referent, headed by a predicate denoting a state that cannot be altered, or that refers to an event in the past. Accordingly, they have to be considered as propositional or speech act modifiers. Chapter 4 is dedicated to the contexts from which epistemic modal verbs are excluded while their circumstantial counterparts are fully grammatical. These non-canonical environments for epistemic modal verbs play an important role in their characterisation. In the present study, the twenty-one most important environments will be thoroughly checked against corpus data. As will be shown, more than half of them are fallacious, as they are attested with epistemic modal verbs. There are only eight environments in which epistemic modal verbs could not be found: (i) They do not occur with verbless directional phrase complements, (ii) they cannot be separated from their infinitive complements in wh-clefts, (iii) they do not undergo nominalisation, (iv) they are exempt from adverbial infinitives

6 | 1 Introduction

and, finally, they cannot be embedded under (v) circumstantial modal verbs, (vi) predicates of desire, (vii) imperative operators or (viii) optative operators. This conclusion is very similar to the one reached by Eide (2005: 9) for Norwegian. As the contexts (iv)–(viii) contribute some sort of circumstantial modal operator, the majority of the non-canonical environments can be accounted for in terms of intervention. Epistemic modal operators cannot occur in the scope of circumstantial modal operators. Furthermore, the corpus study has revealed that the assumption of a distinct ‘objective’ epistemic interpretation, as proposed by Lyons (1977: 799), is misleading. In Chapter 5, it will be shown that reportative uses of wollen and sollen differ substantially from epistemic modal verbs. More specifically, they are more flexible with respect to the contexts in which they can occur. Unlike epistemic modal verbs, they are attested in nominalisations, adverbial infinitives, optatives and embedded under the future auxiliary werden. Furthermore, it will be pointed out that they obtain a different interpretation whenever they are embedded under a past operator. By contrast, it turns out that the so-called evidential verbs scheinen, drohen, versprechen and verheißen belong to a different type of pattern. In Chapter 6, it will be demonstrated how the behaviour of epistemic modal verbs and reportative modal verbs in non-canonical environments can be analysed. The analysis is based on a couple of assumptions. The most basic assumption is that epistemic operators contribute a variable for the deictic centre. In order to be interpreted, this variable needs to be locally bound by an appropriate attitude holder. In the canonical case for epistemic modal verbs, the variable is identified with the most salient referent of the speech act, usually the speaker. While in the case of epistemic modal verbs the variable is only instantiated at the speech act level, the variable is anchored in a very local configuration with reportative modal verbs: it is bound to an argument of the modal verb itself. As there are operators which fail to embed linguistic structures containing unbound variables for the deictic centres, such as circumstantial modal operators, epistemic modal verbs cannot occur in the scope of such operators, since their variable for the deictic centre is left unspecified. By contrast, reportative modal verbs are acceptable in such environments. This explains why reportative modal verbs can occur embedded in configurations in the scope of certain modal operators such as adverbial infinitives or optatives, whereas epistemic modal verbs are banned from such environments. According to this, epistemic modal verbs can be regarded as elements of the category ‘verb’, as their incompatibility with non-canonical environments results from the status of the variable which they introduce. Moreover, it will be demonstrated that approaches in the tradition of Cinque (1999), which analyse modal verbs as functional categories, face serious challenges when intending to account for the data presented here. It appears that only lexicalist ac-

1.4 Results in a nutshell |

7

counts such as HPSG are capable of providing an analysis that is empirically well supported. Finally, the insights from the investigation of the epistemic modal verbs in contemporary language use put us in a position to reconstruct a scenario of how these modifiers came into existence. Chapter 7 is dedicated to the grammaticalisation of epistemic modal verbs. As has been observed by Abraham (1991), Abraham (2001) and Abraham (2005) as well as Leiss (2002), among the earliest uses of epistemic modal verbs there are many that select stative or other imperfective predicates. This can be related to the findings of Chapter 3, where it is shown that circumstantial modal verbs are event modifiers, which are restricted to the modification of predications that involve (Davidsonian) event arguments. Following Maienborn (2003: 106), Kimean state predicates, such as the copula in German, do not contribute an event argument. Yet, in Old and Middle High German instances of circumstantial modal verbs can already be found that embed the copula sein. This situation is unexpected if circumstantial modal verbs are indeed restricted to the selection of predicates that involve an event argument. But as Maienborn (2003: 178, 193) argues, there are two pragmatic repair mechanisms which can render an event argument to a predicate that would otherwise lack such an argument: The temporariness effect, and the agentivity effect. Likewise, Kratzer (1981: 61) argues that there are pragmatic mechanisms of coercion and accommodation that can adapt complements not fulfilling the selectional restriction. As she remarks, this mechanism “is black magic, but it works in many cases”. However, this mechanism is not always easy to detect, especially for L1 learners. In the course of history, there was one generation of L1 learners who were no longer able to decode this repair mechanism. Rather, they reinterpreted the utterance in a more economic way. Since epistemic modal verbs are not restricted to predicates that provide an event argument, they do not require the temporariness effect or the agentivity effect to apply. As circumstantial possibility verbs and epistemic possibility verbs sometimes obtain almost the same communicative effect, this scenario of reinterpretation seems very likely. These results indicate that the grammaticalisation of epistemic modal verbs is, in essence, a change from event modification to clausal modification, and a process which leaves the original category of the grammaticalised element unaffected.

2 Modal verbs: A class struggle Modal verb is probably one of the most common terms in the contemporary description of German grammar, and it is used as if it had always been around. All of the major grammars of German employ this expression, cf. Engel (1996: 463), Zifonun (1997: 1253), Eisenberg (2004: 90), Helbig and Buscha (2001: 114) and Eisenberg et al. (2005). But what exactly does it mean? What is the benefit of using this term? The most common answer would be that this word refers to a group of six particular verbs, which, according to many, constitute a “relatively closed system” and are “part of a grammatical system of rules”, see Buscha, Heinrich and Zoch (1971: 7): (1)

können, müssen, wollen, dürfen, sollen, mögen can must want may shall like

And indeed, this corresponds exactly to what is taught in most schools in countries where German is spoken as a native language, and occasionally in institutions where it is taught as a second language.¹ As this answer is not very precise, further questions arise. What is the particularity of these six verbs? How do they differ from the remaining set of verbs? What makes them so special? Following the traditional view of Vernaleken (1861: 94), Bauer (1850: 102 §166), Curme (1922: 317), Helbig and Buscha (2001) argue that being of preterite present origin, they exhibit a particular morphology, and that they select a bare infinitive and express modality. Accordingly, they constitute a relatively closed group. Furthermore, Helbig and Buscha (2001: 115) point out that modal verbs in German withstand passivisation, and their past participle is always realised with infinitive morphology (IPPeffect, Ersatzinfinitiv). In a similar vein, Griesbach and Schulz (1976: 34) highlight that modal verbs lack imperative morphology. Summing up, in these approaches modal verbs are characterised by morphological anomalies (preterite present origin, lack of an imperative, lack of a passive, IPP-effect), by the selection of a bare infinitive, and by their ability to express modality. Buscha, Heinrich and Zoch (1971) suggest a whole range of further criteria, but they are rather intuitive and do not withstand closer scrutiny. There are a couple of influential studies that are led by these assumptions, in particular that there is a class of modal verbs consisting of these six verbal lexemes. Among others, Bech (1949) and Bech (1951) and Diewald (1999) tried to provide a comprehensive description of the class of elements listed in (1).

1 As Eva Valcheva (pers. commun.) reports, the very same concept of modal verb is taught in schools in Bulgaria as well. https://doi.org/10.1515/9783110411027-002

2.1 Traditional criteria |

9

In the next sections, the criteria suggested by Bech and Diewald will be carefully reviewed. None of them will prove to be reliable enough to justify the assumption of a homogeneous class containing the six items können, müssen, wollen, dürfen, sollen and mögen. Finally, I will arrvive at the conclusion that the term modal verb, as it is most often employed in research on German grammar, is misleading.

2.1 Traditional criteria As discussed above, there is some evidence that the six verbs in (1) constitute a separate class of verbs. What follows is a collection of the phenomena adduced by different proponents of the traditional perspective. This does not mean that all traditionalists agree about the exact number of characteristics of modal verbs. Curme (1922: 317), for instance, only briefly refers to the morphological anomaly of these verbs that is mainly due to their preterite present origin. Other authors, such as Helbig and Buscha (2001) and Griesbach and Schulz (1976), mention further morphological features and, in addition, syntactic peculiarities. It is not always the case that each author was aware of the explanatory power of other potential criteria. The most promising of these potential characteristics will be discussed in the upcoming sections.

2.1.1 Morphological criteria There are two morphological anomalies that can be found among the six traditional modal verbs: an inflectional pattern which is typical of preterite presents and the infinitivus pro participio (IPP-effect). Finally, the lack of an independent imperative form will be discussed. Some authors, e.g. Redder (1984: 305), argue that all of these three features are related to the preterite present origin of the verbs under analysis. 2.1.1.1 Preterite present origins According to Helbig and Buscha (2001: 29), one of the essential properties of the six traditional modal verbs is their preterite present origin. As pointed out by Grimm (1822: 851, 1053), preterite presents are preterite forms of strong verbs that were reinterpreted as independent verbs. Roughly speaking, the accomplishment of an event in the past was reanalysed as a resultative state in present. The case of wissen (‘know’) has been intensively investigated, as is illustrated in Table 2.1. According to Meid (1971: 18), its development originates from an early period of

10 | 2 Modal verbs: A class struggle

Tab. 2.1: Preterite present origin of wissen — the Old High German paradigm

infinitive 1st pers. sg. 2nd pers. sg. 3rd pers. sg. 1st pers. pl. 2nd pers. pl. 3rd pers. pl. meaning

present r¯ıt-an r¯ıt-u r¯ıt-ist r¯ıt-it ¯ r¯ıt-emes r¯ıt-et r¯ıt-ent ‘ride’

past reit-ø reit-ist reit-ø ¯ rit-umes rit-ut rit-un ‘rode’

present (w¯ızz-an) (w¯ızz-u) (w¯ızz-ist) (w¯ızz-it) ¯ (w¯ızz-emes) (w¯ızz-et) (w¯ızz-ent) ‘see’

past (weiz-ø) (weist) (weiz-ø) ¯ (wizz-umes) (wizz-ut) (wizz-un) ‘saw’

⇒ ⇒ ⇒ ⇒ ⇒ ⇒

present wizzan weiz-ø weist weiz-ø ¯ wizz-umes wizz-ut wizz-un ‘know’

Indo-European. Rix (2001: 606) assumes an Indo-European root *ueid ‘see’ with ̑ ̑ its corresponding preterite stems *uóida ‘I saw’ and *uid- ‘We saw’. Already in the ̑ ̑ ̑ Indo-European period, the preterite stems developed an independent meaning. Whereas in the original sense they referred to a seeing event in the past, they refer to a knowing state in the present in its reinterpreted form. Birkmann (1987: 351) illustrates this evolution from Proto-Germanic up to Modern German: (2)

New High German weiß < Old High German weiz < West Germanic *weit < Proto Germanic *wait < Indo-European *uóida ‘I know’ ⇐ ‘I saw’ ̑ ̑ (3) New High German wissen < Old High German wizzum < West Germanic *witum < Proto Germanic *witum < Indo-European *uid- ‘We know’ ⇐ ‘We ̑ saw’ During this process the form wissen maintained its preterite morphology of a strong verb. This becomes visible as soon as it is compared with a preterite form of a verb belonging to the same class of ablaut, as reiten (‘ride’), for instance. And indeed, following Birkmann (1987: 135) and Braune and Reiffenstein (2004), wissen inflected for present tense behaves exactly as reiten in its preterite use, as is illustrated for Old High German in Table 2.1. Even if Pokorny (1959: 1126) adduces a verb w¯ızzan ‘look out, observe’ for Old High German, this does not mean that the process of reinterpretation only took place in that period. Effectively, the emancipation of the new meaning of wizzan already took place in Indo-European times. The reason why w¯ızzan is nevertheless included in Table 2.1 is only for ease of illustration. It only demonstrates what the original stem *ueid would have ̑ ̑ looked like in Old High German. In essence, these patterns remain the same for New High German. There are three characteristics that are particular to preterite presents: (i) the 1st and the 3rd person singular remain without suffix in present tense. A similar observation was made by Claius (1578: 96), who noticed that there are nine verbs

2.1 Traditional criteria |

11

that lack suffixes in the 1st and 3rd person singular, which makes them to appear monosyllabic: können, mögen, woellen, sollen, wissen, taugen ‘to be good for sth’, thuerren ‘dare’, düerfen and müssen. (ii) They involve a vowel alternation between the present tense indicative stems in singular and plural, and (iii) they exhibit a further vowel alternation between the stem of the present tense and the past tense. As for the six verbs listed in (1), it turns out that indeed almost all of them are of preterite present origin. As, among others, Braune and Reiffenstein (2004) illustrate, können, müssen, dürfen, sollen and mögen can be derived from preterite stems of other verbs. Based on the observations about the Gothic counterpart viljan ‘want’, Grimm (1822: 853) illustrates that wollen originates in a subjunctive of the past form of a volitional verb. A similar analysis of the development of German wollen was suggested by Braune (1886: 259). However, in the course of history, wollen assimilated its morphological properties according to the paradigm of preterite presents, as illustrated by Braune and Reiffenstein (2004). The different origin of wollen is partly reflected in its deviating inflectional pattern. It does not involve a vowel alternation between the preterite stem and the infinitive. Therefore criterion (iii) for preterite present is not met, as indicated in Table 2.2. Upon closer inspection, however, it turns out that the genuine preterite present sollen even fails to fulfil two of the characteristics particular to preterite presents. It involves no vowel alternation at all; correspondingly, criteria (ii) and (iii) are not met. In a similar fashion, muozan lacked the vowel alternation between the indicative forms in singular and plural in Old High German, violating criterion (ii), as illustrated by Birkmann (1987: 129). Finally, the status of criterion (iii) is unclear, as it does not uniquely apply to preterite present verbs, but can be found with many more verbs. More specifically, the vowel alternation between the infinitive and the past tense stem is a characteristic that affects most of the irregular verbs as well. As illustrated by Eisenberg et al. (2005: 491–502), there are more than 190 irregular verbs that display a vowel alternation between the present stem and the past stem. As a consequence, preterite present morphology cannot be regarded as a suitable property to unify the traditional six modal verbs in a homogeneous class. A definition of the modal verbs based on the preterite present morphology faces a further challenge, since it incorrectly includes wissen, which is apparently the oldest among the preterite presents and, unlike sollen, has preserved all of the relevant features of preterite presents. Typically, authors who suggest a definition of modal verbs which is based on their preterite present origins, Curme (1922: 317), acknowledge at some later point that wollen has, in fact, a different origin and only assimilated over the course of time. In this respect, German behaves differently from English, where the class of preterite presents coincided with a group of verbs with ‘modal’ meanings, as

12 | 2 Modal verbs: A class struggle

Tab. 2.2: Modal verbs and preterite present morphology

können müssen mögen dürfen wollen sollen wissen

1st and 3rd p. present tense without suffix 1st /3rd p. sg. kann-ø muß-ø mag-ø darf-ø will-ø soll-ø weiß-ø

vowel change 1st –3rd p. sg kann muß mag darf will soll weiß

1– 3rd p. pl. können müssen mögen dürfen wollen sollen wissen

vowel change infinitive können müssen mögen dürfen wollen sollen wissen

past tense konnte mußte mochte durfte wollte sollte wußte

Lightfoot (1979: 102) has pointed out. All of the other preterite presents vanished. This in turn triggered a radical process of syntactic change with the result that all of the preterite presents were reanalysed as auxiliaries. Lightfoot (1979: 98) stresses that preterite presents in Old English sculan, willan, magan, cunnan and motan exhibited all features that are typical of a canonical verb. In the 16th century, however, they suddenly lost these features and were reanalysed as functional elements. To sum up, preterite present morphology cannot be employed as class defining property to separate the six traditional modal verbs from the remaining verbal elements: wollen is not a preterite present and there is a further preterite present, wissen, which is usually not considered as a modal verb. 2.1.1.2 The IPP-effect: The Ersatzinfinitiv A further criterion that is invoked in traditional definitions is the Ersatzinfinitiv or infinitivus pro participio (IPP-effect), as in Helbig and Buscha (2001: 115). Verbs such as dürfen will usually be realised with infinitive morphology whenever they are embedded by the perfect auxiliary haben ‘have’. As opposed to the canonical type of verb, the ge-participle is not available in this environment for the six traditional modal verbs.² This holds true at least in Standard German, while some Western German dialects do not exhibit the IPP-effect and employ the ge-participle in corresponding contexts.³ 2 The status of the IPP is fairly contested. Some scholars, e.g. Hinterhölzl (2009: 198), argue that it is a genuine infinitive, others argue that it is a hidden participle. A detailed discussion is provided in Hinterhölzl (2009: 197–198). As it fulfils a similar role as the common ge-participle, it will be glossed as: ppp(ipp). However, this should not be taken to imply that the IPP has really substantially the same function as a ge-participle. 3 In some varieties, the acceptability of ge-participles increases when the bare infinitive appears separated from the modal verb. However, as the following sentences were taken from a show of

2.1 Traditional criteria |

(4)

13

Das hättest du nicht sagen dürfen / *gedurft that had you neg say-inf may-ppp(ipp) may-ppp(ge) ‘You shouldn’t have said that.’

The relation to the present preterite history of these verbs is obvious. Being former preterite tense forms of some other verbs, the traditional six modal verbs were lacking a full inflectional paradigm. Therefore, it became necessary to develop a past participle of their own. But as already explicitly pointed out by Kurrelmeyer (1910: 167), the IPP-effect is not a genuine innovation of modal auxiliary verbs: The first of the traditional modal verbs that can be found with the IPP-effect is müezen in the 15th century, and the last one is sollen, which is only recorded from the 16th century onwards. Some other verbs exhibit the IPP-effect much earlier: tun ‘do’ (1259), helfen ‘help’ (1263), hoeren ‘hear’ (13th century), heizen ‘command’ (1277), lazen ‘let’ (13th century), sehen ‘see’ (14th century), machen (1475). In a similar vein, Hinterhölzl (2009: 202) argues that the IPP-effect originally emerged with heißen, lâzen, tun and hoeren and only spread to the preterite presents over the course of time. This is also confirmed by Ebert et al. (1993: 413–414), who show that müssen already occurred with the IPP-effect in the 13th century, whereas the remaining traditional modal verbs wollen, mögen and können only acquired it in the course of the 15th century, or even later, such as sollen and dürfen. As already mentioned above, it seems plausible that preterite present verbs and other verbs with defective paradigms, such as wollen, seek to complete their morphological inventory. The remaining preterite presents, which are not part of the traditional six modal verbs, are also found with the IPP-effect, at least in earlier stages of German. Kurrelmeyer (1910: 164) gives an example for türren ‘dare’ with an infinitive complement displaying the IPP-effect from the year 1375. Moreover, there are numerous occurrences of wissen with a zu-infinitive from the 17th century that display an interpretation which refers to a mental ability read-

the comedian Karl Valentin, conclusions concerning linguistic theories should be handled with care. (1) Wollen hätte ich schon gedurft... want-inf had I though may-ppp(ge) ‘It was okay for me to want it.’ DeReKo: M09/AUG.63846 Mannheimer Morgen, 15.08. 2009. (2)

Wollen hätten wir schon mögen, aber trauen haben wir uns nicht gedurft. want-inf had we though like-ppp(ipp) but dare-inf had we us neg may-ppp(ge) ‘To want it was appealing, but we were not permitted to dare it.’ DeReKo: NUN08/JUL.00977 Nürnberger Nachrichten, 09.07. 2008.

14 | 2 Modal verbs: A class struggle

ing exhibit the IPP-effect, as illustrated in examples (5)–(12). This was already pointed out in the 17th century by Bödiker (1698: 109; see Section 2.3 for more details), as well as Grimm (1837: 168), Sanders (1908: 428), Alban (1992: 6), Ebert et al. (1993: 413) and Maché and Abraham (2011: 256). In contrast to the remaining preterite presents, wissen is persistently used with an infinitive with zu. (5)

der Arzt Asclepiades hat durch den beweglichen Wollaut the physician Asclepiades has by the moving euphony lieblich=zusammenklingenden Seiten/ die abweichende der chords the deviating the-gen lovely.harmonic Vernunft abzuhalten wissen/⁴ reason detain-inf know-ppp(ipp) ‘The physician Asclepiades knew how to use the moving euphony of harmonic chords to prevent distraction of thought.’

(6)

liesz mein buechlin, so wirstu sehen, das der luegengeist nicht read my book so will.you see that the lye.spirit neg hat wissen zu antworten⁵ has know-ppp(ipp) zu answer-inf ‘Read my book and you will see that the lying spirit was not able to answer.’

(7)

Agricola: Ich hab nichts darinn wissen zuo meiden Agricola I have nothing there.in know-ppp(ipp) zu avoid-inf oder außzelassen.⁶ or zu.ignore-inf ‘Agricola: I could not have avoided or ignored any of them.’

(8)

Er hat sich wissen ò gewust in seine Gnade he has ana know-ppp(ipp) or[Italian] know-ppp(ge) in his mery einzuschleichen⁷ zu.-inf ‘He made himself endear to him.’

(9)

Sie hat nicht wissen ò gewust zu hüten⁸ She has neg know-ppp(ipp) or[Italian] know-ppp(ge) zu watch-inf ‘She couldn’t watch (it).’

4 Schottelius, Ausführliche Arbeit von der Teutschen HaubtSprache (1663), p. 67. 5 Martin Luther 26, 613 W, [as cited in Grimm DWB]. 6 Johannes Cochläus, Ein heimlich Gespraech von der Tragedia Johannis Hussen, Actus tertii scena unica, (1538), B 3a. 7 Matthias Kramer, Das herrlich-Grosse Teutsch-Italiänische Dictionarium (1702), p. 1368. 8 Matthias Kramer, Das herrlich-Grosse Teutsch-Italiänische Dictionarium, (1702), p. 1368.

2.1 Traditional criteria |

(10)

15

er hat es nicht auszurichten wissen⁹ He has it neg transmit.zu-inf know-ppp(ipp) ‘He could not transmit it.’

(11)

Sie hätten damit nichts wissen anzufangen¹⁰ they had with.it nothing know-ppp(ipp) start.zu-inf ‘They did not know what to do with it.’

(12)

Hat Rom sein siebenbergigt Haupt sonst nirgends hin zulegen Has Rome its seven.hilled head apart nowhere par lay-inf wissen¹¹ know-ppp(ipp) ‘Rome did not have any other place for its head made of seven hills to lay.’

Yet, there are instances of wissen which do not carry the infinitive particle zu and which govern a bare infinitive complement. This behavior may be caused by the conjunction with a more prototypical modal verb, wollen, which is restricted to the subcategorization of bare infinitive complements. But the fact that wissen occurs conjoined with woellen sharing the same infinitival complement illustrates how close these verbs are syntactically speaking. (13)

Das Ebreisch wort Moed / hab¯e wir nicht anders the Hebrew word Moed have-1.P.Pl we neg differently wissen noch woellen deudschen¹² know-ppp(ipp) nor want-ppp(ipp) germanise-inf ‘We were not able nor did we want to translate the Hebrew word Moed in a different way.’

Interestingly, most of the occurrences of wissen collected here are in the scope of negation. This is reminiscent of the negative polar behaviour of other modal verbs such as the raising pattern of wollen, the emotive use of mögen and earlier uses of dürfen and brauchen, as is illustrated in Sections 2.2.3, 2.2.4, 2.2.7 and 2.2.9. Moreover, these examples exhibit a word order that contradicts the patterns typical of coherence/clause union (2-1 or 1-3-2). Only in example (10) given by Grimm (1837: 168), does the subcategorised infinitive complement immediately precede wissen. As Grimm (1837: 168) notices, the IPP-effect with wissen can be frequently observed in the vernacular. Nevertheless, he regards this use as false, since the

9 As discussed in Grimm IV, 168. 10 Hebel, [as cited in Sanders (1908), p. 428]. 11 Daniel Casper von Lohenstein, Ibrahim Bassa, (1653), [as cited in Schoetensack 1856, 298]. 12 Martin Luther, Biblia, Mose, Das Ander Buch, XXVII, footnote a, p. 53 (1533).

16 | 2 Modal verbs: A class struggle

IPP-effect is a property that is restricted to verbs that sub-categorise bare infinitive complements. Instead, he recommends employing the ge-participle gewußt in these contexts. As demonstrated by Maché and Abraham (2011: 269), there are at least two properties that make verbs with non-finite complements susceptible to the IPPeffect: a defective paradigm, as in the case of preterite presents and wollen, and raising infinitives, such as the subject-to-object raising (AcI) verbs lassen ‘let’, hören ‘hear’, sehen ‘see’ and fühlen; ‘feel’ and the subject-to-subject raising verbs, pflegen ‘used to’ and düncken ‘seem’, which both exhibited the IPP-effect during the Early New High German period. A third relevant property is the selection of bare infinitive complements. Returning to the modal verbs, it becomes clear why they are such prominent exponents of verbs with the IPP-effect: They carry all of those properties. They exhibit raising patterns (as will be shown in more detail in Section 2.2), a defective paradigm, and finally, they select bare infinitive complements. This explains why they are susceptible to this morphological anomaly to such a great extent. At this point, it also becomes clear that the traditional six modal verbs did not grammaticalise as a block; rather, each verb had its own development and each development had its own pace. This in turn demonstrates that the extension of the group of verbs with auxiliary-like behaviour differed with respect to the particular period. As will be shown in Section 2.1.2.1, each of the traditional six modal verbs developed the ability to select bare infinitive complements at an individual point of time. If there were periods during which the traditional modal verbs did not constitute a homogeneous class, one may have to reassess the empirical evidence in order to find out whether there is a period at all in which these six verbs form a class of this type. Even if the six traditional modal verbs acquired the IPP-effect before they developed a ge-participle of their own, as Ebert et al. (1993: 414) shows, it turns out that, from a diachronic perspective, the IPP-effect is not a property which is restricted to the six traditional modal verbs.¹³

13 Kurrelmeyer (1910: 165) discusses a somewhat controversial example from a charter from the year 1332, which is taken to be a ge-less past participle: (1) Swelhie fraw niht gehorsam hat getan oder tun wolt which.ever lady neg obedience has do-ppp(ge) or do-inf may-ppp(?) ‘Whatever lady that refused to obey or wanted to do so.’ Arguably, wolt could also be analysed as preterite 3rd person singular form. Yet, Schallert (2014: Sect. 4.1, 2014: 187–188) and Fleischer and Schallert (2011: 184) have pointed out that the participle forms of preterite-presents were occasionally realised as weak participle lacking the geprefix, which are commonly refered to as ‘truncated participles’.

2.1 Traditional criteria |

17

Shifting to a synchronic view, the situation is no different. Apart from the six traditional modal verbs, there is at least one further verb that obligatorily exhibits the IPP-effect whenever it is embedded by the perfect auxiliary haben: The causative use of lassen ‘let’, as has been pointed out by Schmid (2000: 328). Once more, the ge-participle is not available, as in example (14). Therefore, beginning with Becker (1841: 219), lassen has sometimes been counted among the traditional class of modal verbs.

Sie hat ihren Mann umbringen lassen / *gelassen She has her husband kill-inf let-ppp(ipp) let-ppp(ge)

(14)

‘She let him be killed.’

Note that lassen also has a permissive use (‘to tolerate’) and a relinquative one (‘to leave something behind, let go’), as argued by Maché and Abraham (2011: 260). According to Aldenhoff (1962: 204), the causative and the permissive use always exhibit the IPP-effect, whereas the relinquative use is optionally realised as the geparticiple. Some speakers, however, also accept ge-participles of permissive lassen. Finally, the remarkable case of brauchen ‘need’ has to be mentioned, which, in contemporary standard German, always exhibits the IPP-effect. Again, the geparticiple is ungrammatical: (15)

a.

Aber Flavio Cotti hätte nicht zu kommen But Flavio Cotti have-sbjv.pst neg to come-inf brauchen.¹⁴ need-ppp(ipp) ‘But it wouldn’t have been necessary for Flavio Cotti to come’

b.

* Aber Flavio Cotti hätte nicht zu kommen gebraucht.

Being very close to the traditional modal müssen in semantic respect, brauchen, too, seems to have assimilated to its counterpart in morphological respects. Most importantly, this concerns the development of the IPP-morphology. It was already observed by Grimm (1837: 168, 949) that brauchen occasionally exhibits the IPPeffectr, as is shown in his own example (16).¹⁵

14 DeReKo: E98/JUN.15388 Zürcher Tagesanzeiger, 19/06/1998. 15 As will be shown in Section 2.2.9 in more detail, in some regions, brauchen is even subject to further processes of assimilation. According to André Meinunger (pers. commun.), this morphological assimilation of brauchen towards the “modal” morphology is even more developed in the region around Wuppertal, where speaker omit the t-suffix of the 3rd person indicative singular, as in the sentence Er brauch-ø nicht kommen ‘He need-ø not come’. Similar observations about brauchen have been already made by Wurzel (1984: 117 & 149), Birkmann (1987: 5) and Girnth (2000: 115) and Beringer (s.a.). In this respect, brauchen is reminiscent of need in Modern Eng-

18 | 2 Modal verbs: A class struggle

(16)

das hätte ich nicht zu tun brauchen (gebraucht) that have-sbjv.pst I neg to do-inf need-ppp(ipp) need-ppp(ge) ‘I wouldn’t have had to do it.’

Even if he acknowledges that this pattern is frequent in colloquial speech, Grimm is reluctant to consider it as fully grammatical. According to him, the IPP-effect only occurs with bare infinitives. In contrast, brauchen sub-categorises for a zuinfinitive. For this reason, he refers to the correct alternative, the ge-participle, in brackets. As Sanders (1908: 101) demonstrates, brauchen with an infinitive complement could instead be realised as a ge-participle up to the 19th century: (17)

er hätte nur die Regungen der eigenen Brust zu he have-sbjv.pst only the emotions the-gen own chest to besingen gebraucht¹⁶ sing need-ppp(ipp) ‘He only needed to sing about the emotions in his chest.’

In opposition to Grimm (1837), Sanders (1908: 101) considers brauchen with the IPP-effect as grammatical. Moreover, he argues that the infinitival particle zu can occasionally be dropped, which is remarkable since Sanders takes a rather normative perspective. The optionality of the zu-particle will be dealt with in Section 2.2.9.4. Apart from causative lassen and brauchen with an infinitive, there is a large group of verbs that optionally permit the IPP-effect: Following Schmid (2000: 330) in particular, this concerns the AcI verbs (object-to-object raising, exceptional case marking) sehen ‘see’, hören ‘hear’, fühlen ‘feel’, and benefactive verbs such as helfen ‘help’, lernen ‘learn’ and lehren ‘teach’. Aldenhoff (1962) and Sanders (1908: 222) provide an extensive discussion of this issue. In a less systematic way, Heyse (1822: 413) has already observed that the IPP occurs with a whole range of verbs: dürfen, heißen, helfen, hören, können, lassen, mögen, müssen, sollen, sehen, wollen, lehren and lernen. Yet, Heyse (1822: 414) argues that this use is a severe violation of the logical principles (‘grober Verstoß gegen die Logik’). Accordingly, he suggests to better use the ge-participles of these verbs, even if they take infinitive complements. Likewise, Schoetensack (1856: 298) has pointed out that the IPP has been observed with a similar group of verbs: hören, heißen, sehen, helfen, lassen, sollen, wollen, mögen, dürfen, müssen, wissen, können, fühlen, lehren and lernen.

lish, which lacks an s-suffix if it is used with an infinitive complement, as has been described by Sweet (1891: 425). 16 Heine 2, 307, as cited in Sanders (1908: 101).

2.1 Traditional criteria |

19

As demonstrated above, the IPP-effect is not a property that is restricted to the traditional six modal verbs. Hence, it is not suitable as class defining property. For the sake of completeness, note that some grammarians indeed suggest that the IPP-effect is the essential criterion for auxiliary-hood, acknowledging that the extension of such a class does not exactly correspond to the six traditional modal verbs. The first person who discussed the IPP-effect was Ölinger (1574: 151). As he observes, the five verbs woellen, sollen, doerffen, koennen and moegen do not employ a ge-participle, but rather an infinitive, whenever they are embedded under a perfect tense auxiliary. Bödiker (1698: 109), in turn, argues that, because of their morphological anomaly, these five verbs together with muessen and wissen constitute an auxiliary-like class. In a similar vein, Sanders (1908: 222) argues that there are a number of auxiliaries in German characterised by the IPP-effectand encompassing the following items: d"urfen ‘may’, heißen ‘command’, helfen ‘help’, hören ‘hear’, können ‘can’, lassen ‘let’, lehren ‘teach’, lernen ‘learn’, machen ‘make’, mögen ‘like’, müssen ‘must’, sehen ‘see’, sollen ‘shall’, wollen ‘want’ and occasionally brauchen ‘need’, pflegen ‘used to’, suchen ‘seek’, rare empfinden ‘feel’, erblicken ‘see’, finden ‘find’, fühlen ‘feel’, schauen ‘look’, wissen ‘know’, and zeigen ‘show’. In any case, the IPP-effectis not a property that justifies a modal verb class in the traditional extension, neither from a synchronic nor from a diachronic perspective. Nevertheless, it has proven to be a powerful criterion since there are only two more verbs apart from the traditional six modal verbs that obligatorily exhibit the IPP-effect: (the causative use of) lassen and brauchen. 2.1.1.3 Imperative Some authors, e.g. Welke (1965: 14), Eisenberg (2004: 91) and Erb (2001: 97), argue that the six traditional modal verbs are further characterised by their inability to form imperatives. This perspective has already been taken by Claius (1578: 103), who claimed that those verbs today referred to as preterite presents, with the exception of wissen, do not have an imperative. In a similar fashion, Adelung (1801: 1608) argues that wollen does not form an imperative. It is not evident whether these observations indeed hold, since at least two hundred years later the imperative of wollen is documented, as is illustrated by the dialogue below taken from Goethe’s Faust (cf. 18). Some authors, such as Voß, use the imperative even when wollen occurs with infinitive complement (cf. 19).¹⁷

17 Johann Wolfgang von Goethe, Faust - der Tragödie erster Teil, V 4543, (1808).

20 | 2 Modal verbs: A class struggle

(18)

a.

MARGARETE: [...] Du gehst nun fort? Oh Heinrich, könnt ich you go par away oh Heinrich could I mit! with

b.

FAUST: Du kannst! So wolle nur! Die Tür steht offen! you can so want-imp just the door stands open

(19) Woll’ auch diesen verzeihn! – Für uns nicht brauchst du zu

want also them forgive-inf beten!¹⁸ pray-inf

for us neg need

you to

‘Forgive them, too – You do not need to pray for us.’

As already pointed out by Zifonun (1997: 1254), Hetland and Vater (2008: 99) and Vater (2010: 108) wollen has an imperative. Admittedly, this form is only available when wollen is used without an infinitive complement. It is important to keep in mind that even if the imperative usage of wollen is rather rare, it is much more acceptable than the imperative usage of other modals. This illustrates that there is a substantial difference between wollen on the one hand and the remaining traditional modal verbs on the other hand. In a similar vein, Hetland and Vater (2008: 97) observe that each modal behaves differently in a morpho-syntactic manner. The markedness of the imperative with the traditional six modal verbs might be also related to the defective nature of their paradigm. Likewise, the imperative of the last remaining preterite present, apart from the modal verbs, wissen, is equally marked as the one of wollen, at least in Contemporary German. As already pointed out by Claius (1578: 103), the lack of imperative forms is a criterion that holds for most preterite presents, including verbs that do not belong to the modal verb class in its traditional extension, such as thar ‘dare’ and taug ‘suit’. Accordingly, this criterion does not justify treating the six traditional modal verbs as a homogeneous class, either.

2.1.2 Syntactic criteria The most important syntactic criterion that is invoked for the separation of the six traditional modal verbs from the remaining elements of the verbal category concerns the category of the complement.

18 Verwandlungen, Third book – Pentheus, I, 192.104, translated by Johann Heinrich Voß, Berlin: Friedrich Vieweg der Ältere (1798).

2.1 Traditional criteria |

21

2.1.2.1 The sub-categorisation of bare infinitive complements As Welke (1965: 11 & 22) and Griesbach and Schulz (1960: 41 & 65) argue, an essential characteristic of the traditional six modal auxiliaries is the selection of bare infinitive complements. But as Welke (1965) acknowledges, there are many more verbs in Contemporary German that sub-categorise bare infinitive complements. Following Maché and Abraham (2011: 236), at least ten different groups of predicates come into consideration. On the one hand, there are verbs that take bare infinitive complements but never zu-infinitives (cf. 20):¹⁹ the ‘do-support auxiliary’ (cf. 20a), the future auxiliary (cf. 20b), the subjunctive auxiliary (cf. 20c), the traditional modal verbs (cf. 20d), subject-to-object raising (AcI) verbs (cf. 20e), verbs of motion (cf. 20f), verbs of caused motion (cf. 20g), and durative verbs (cf. 20h).²⁰ On the other hand, there are verbs for which both types of realisation of the non-finite complements can be found: bare infinitives and zu-infinitives, see (21). Above all, this concerns benefactive verbs (cf. 21a and 21b). The latter is a somewhat remarkable case because the realisation of the complement type is governed by the register. Whereas zu-infinitive prevails in written standard language, the bare infinitive is almost restricted to spoken language. (20)

a. tun ‘do’ b. wird fut.aux c. würde sbjv.aux d. können, müssen, wollen, dürfen, sollen, mögen, (möchte/n) e. sehen ‘see’, hören ‘hear’, fühlen ‘feel’, finden ‘find’, spüren ‘feel’, lassen ‘let’, heißen ‘command’, machen ‘make’, haben ‘have’ f. kommen ‘come’, gehen ‘go’, fahren ‘ride’ g. schicken ‘send’, senden ‘send’ h. bleiben ‘stay’, sein ‘be’

(21)

a. brauchen b. lernen, helfen, lehren

This classification does not entirely correspond to the one of Askedal (1989: 5). He suggests that zu-infinitives occurring with the verbs of motion in example (20f)

19 Of course, some of the verbs below, such as sein ‘be’ or haben ‘to have’ can be found with the zu-infinitive or other types of non-finite complements, but in these instances they will exhibit a different semantic interpretation. 20 As has been pointed out by Langer (2001: 63), the auxiliary tun in German has a whole range of functions: It can bear the past or subjunctive of the past morpheme and it is used to obtain particular configurations of information structure such as V-topicalisation.

22 | 2 Modal verbs: A class struggle

and verbs of caused motion in example (20g) have to be considered as their complements. Yet, he admits that the option of containing the zu-infinitive in example (22b) is only rarely used and is hardly found in verbal complex configurations (“obligatorily coherent” in the terms of Bech (1955)), where the infinitive complement has to precede the finite verb, as in example (22a). (22)

a.

obwohl sie jede Woche zum Priester [beichten ging] though she every week to.the priest confess-inf went ‘Although she went to the priest to attend her confession every week.’

b.

obwohl sie jede Woche zum Priester [ging] [zu beichten] though she every week to.the priest went zu confess-inf ‘Although she went to the priest each week in order to attend her confession.’

Therefore, it does not seem plausible that the two types of infinitives have the same status. And there are more arguments against the hypothesis considered by Askedal (1989). Whereas the goal PP zum Priester ‘to the priest’ can be omitted in the first example without any ado, the omission of the goal PP is subject to many more restrictions in the second example, indicating that the bare infinitive may only function as a goal argument. Moreover, the first pattern only entails the realisation of the event expressed by the infinitive in the examples above. Whereas the bare infinitive typically encodes the goal of the movement, the zu-infinitive rather indicates its purpose. Accordingly, the latter could easily be identified as reduced forms of adverbial um-zu-infinitives, which express the purpose of the event described in the main clause, as discussed by Eisenberg (1992, 2004: 351). This illustrates, once again, that bare infinitive complements are found with a considerable number of verbs in Contemporary German. However, focusing on verbs where the infinitival subject is co-referential with the matrix subject, Welke (1965: 11 & 22) and Zifonun (1997: 1253-4) argue that most verbs of this class are restricted to a small group of types of infinitives. In particular, Welke mentions verbs of motion, such as kommen and gehen, and the durative verb bleiben, which is restricted to the selection of stative predicates. According to them, the only group of verbs which does not exhibit selectional restrictions with respect to the infinitive of this type encompasses the six traditional modal verbs. Furthermore, there are two more verbs that behave accordingly, but Welke (1965: 11) explicitly excludes both of them from his definition: The auxiliary tun for not belonging to the standard variety, and werden for the lack of past forms. As he acknowledges himself, his approach is somewhat arbitrary. Note that Welke’s observation above is not quite correct, as even the six traditional modal verbs fail to embed certain types of stative predicates (individual level predicates) in their non-epistemic variant, as will be shown in more detail in Section 3.2. Moreover, Engel (1996: 476) argues that tun and bleiben do not belong to the class of modal verbs since they use -t in

2.1 Traditional criteria |

23

third person singular, and as opposed to modals they never embed an infinitive perfect (* Er tut geschlafen haben ‘He does have slept’). As it turns out, the sub-categorisation of bare infinitive complements is not a property that is unique to the six traditional modal verbs. As a consequence, it cannot serve to justify a syntactically homogeneous modal verb class in the traditional extension. Nevertheless, it proves to be a powerful criterion that almost manages to separate the six traditional modals from the remaining verbs in German. This will be illustrated in Section 2.1.4. Assuming that the selection of a bare infinitive complement is the main characteristic of modal auxiliary-hood, the situation is once again different in earlier stages in German. As pointed out by Demske (2001: 76), most verbs that take nonfinite complements in Old High German are not restricted to a particular type of infinitive. Without that, the interpretation is affected and they may either take bare infinitive complements or zu-infinitives. As Demske (2001: 74) stresses, a small group of verbs is only recorded with bare infinitive complements and never with zu-infinitives: The preterite presents kunnan ‘be mentally able to’, durfan ‘need’, scolan ‘shall’, mugan ‘can’, muoz ‘to have space’, gitar ‘dare’ and the verb wellen ‘want’, the perception verbs hôren ‘hear’ and sehan ‘see’, causative verbs lâzan ‘let’, heizan ‘command’ and gituon ‘do, make somebody do something’, and finally the raising verbs scînan ‘seem’ and thunken ‘seem’.²¹ Birkmann (1987) takes a different perspective on the situation in Old High German. In contrast to Demske (2001), his study is restricted to preterite present verbs. According to his investigation of the Isidor (late 9th century) and a couple of smaller texts, skulan, *muozan, eigan ’have’, magan, kunnan can be considered as auxiliaries since they occur with bare infinitive complements; kunnan is additionally used as a main verb. In contrast, Birkmann (1987: 161) argues that wizzan and thurfan lack a use as an auxiliary and only occur as main verbs. Since Demske (2001) does not give precise examples for most of the preterite presents she discusses, it is not entirely clear how to cope with the minor contradictions between her observations and the ones made by Birkmann (1987). In any case, Birkmann (1987: 144) demonstrates that kunnan was not frequently used until Notker in the early 11th century, and that the situation for thurfan is similar. Diewald (1999: 297) only considers sculan, mugan and wellen to be sufficiently grammaticalised in Old High German. Accordingly, she argues that these are the only modal verbs in that particular period. She explicitly excludes thurfan and kunnan since she has only found occurrences with nominal complements in her in-

21 As Birkmann (1987: 155) demonstrates, the entire paradigm of the verb muozan cannot be found in Old High German. For this reason, only the finite form is mentions here.

24 | 2 Modal verbs: A class struggle

vestigation, but none with an infinitive. Moreover, Diewald (1999: 299) does not regard muozan as a modal verb because it does not exhibit a ‘modal semantics’, even though it is attested with infinitival complements. As she argues, the meaning of muozan from that time is to be paraphrased as ‘to have space to do something’. Her approach, however, is controversial. First of all, it is not clear why Diewald (1999: 299) treats kunnan in Old High German as a main verb, as she explicitly refers to Birkmann (1987), who demonstrates that kunnan also occurs with an infinitive in that very period. Moreover, her notion of ‘modal semantics’ is rather intuitive and not well defined. In any case, Birkmann (1987: 144) and Diewald (1999: 299) agree that können and dürfen with an infinitive complement are rare until the end of the Middle High German period. As has been seen, even from a diachronic perspective, the selection of bare infinitives is not a feature that distinguishes the six traditional modal verbs from the remaining verbs. Rather, some members of the traditional group, such as thurfan and kunnan, are hardly observed with an infinitive for the Old High German period. Moreover, Birkmann (1987: 144) and Demske (2001: 74) demonstrate that there are also additional verbs in that period that obligatorily select bare infinitives, such as the preterite presents eigan ‘have to’ and gitar ‘dare’, subject-to-subject raising verbs and subject-to-object raising verbs. As in Contemporary German, the number of verbs that are sub-categorised for bare infinitives is fairly restricted in Old High German. The two stages differ significantly with respect to the particular verbs that belong to this pattern. But in none of the periods investigated so far does the group of verbs which select bare infinitives correspond exactly to the traditional six modal verbs. Moreover, it turns out that each of the traditional modals acquired the ability to select bare infinitives at a different historical period. As already shown in Section 2.1.1.2, each verb has its own development and each development its own pace. Correspondingly, there is no logical necessity for the class of modal verbs in German to encompass those six members that it encompasses. In fact, the opposite is true, as there is a significant amount of evidence that there might never have been a discrete class of modal verbs, but rather, a loose compound that is in constant change. In a similar spirit, Wurzel (1984: 149) argues that, from the outset, there were two different classes: preterite presents, and verbs that select bare infinitives with modal meanings. Crucially, those classes partially overlapped. Over the course of history, the two classes became increasingly congruent: The non-modal preterite presents lost their anomaly and the non-preterite presents with modal meanings such as wollen and brauchen assimilated to the preterite present morphology. Summing up, the ability to select bare infinitive complements is not restricted to the six traditional modal verbs, neither synchronically nor diachronically. In Contemporary German, at least two verbs behave in a comparable way with re-

2.1 Traditional criteria |

25

spect to sub-categorisation, werden and brauchen in the spoken language. Both of them select bare infinitive complements. Since this property is restricted to the six traditional modal verbs and a small group of verbs apart from them, it appears to be suitable as a class defining property. In this case, however, it would be necessary to reassess the extension of the class, as will be discussed in Section 2.1.4. 2.1.2.2 Subject-to-subject raising Recently, another syntactic property has attracted much attention in the discussion about the characteristics of modal verbs in German. Öhlschläger (1989) and Wurmbrand (1999) and Wurmbrand (2001) argue that modal verbs in German are subject-to-subject raising verbs throughout – with the exception of wollen, and the ability interpretation of können (as well as möchte, which is analysed by both authors as an independent lexical item). These are verbs that lack a subject argument of their own and raise their subject from the embedded infinitive. Wurmbrand (2001: 187) subsumes all modal verbs with raising patterns under the proper syntactic category Mod0 . In their epistemic interpretation, they are generated as a higher category in Aux0 . However, there are subject-to-subject raising verbs apart from the six traditional modal verbs, such as scheinen ‘seem’, drohen ‘threaten’, versprechen ‘promise’, and pflegen ‘be wont to’. Wurmbrand (2001: 205) argues that all of them can be analysed as epistemic modal verbs. As will be shown in Section 2.2, these four raising verbs differ from epistemic modal verbs in crucial respects and therefore need to be treated separately. In a recent study, Gergel and Hartmann (2009: 327) suggest in a radical manner that even the apparent control verb wollen needs to be considered as a raising verb. Since their Generalized Raising Hypothesis is based upon some non-trivial and theory immanent assumptions such as raising into theta positions, the discussion will be suspended here and resumed in the case studies of control an raising verbs in Section 2.2. Crucially, a notion of modal verbs that is based on raising will not obtain a class extension corresponding to the six traditional items.

2.1.3 Semantic criteria Finally, and most notably, most modal verb definitions also involve a semantic dimension. This is not surprising since the term modal already refers to a semantic phenomenon. But as this definition does not concern the material form of a sign but its immaterial function, there is not so much consensus on what the essential

26 | 2 Modal verbs: A class struggle

semantic criterion is for modal verbs. In the upcoming sections, the most plausible criteria will be briefly reviewed. 2.1.3.1 Modality Like many others, Helbig and Buscha (2001: 44) assume that the traditional six modal verbs are characterised by the fact that they express a modality. According to them, a modality can be realised as an ability, a possibility, a necessity, a wish or the attitude of the speaker. Even if these notions intuitively share some common properties, it is not a trivial matter to identify them. All of these expressions locate the event or state denoted by the verb in some idealised worlds that are distinct from our world. Therefore, a modalised event need not be realised in the actual real world. Portner (2009: 1) suggests a similar definition: ‘Modality is the linguistic phenomenon whereby grammar allows one to say things about, or on the basis of, situations which need not be real’. But as already pointed out by Welke (1965: 19), such an attempt of a semantic definition that is based on the expression of a modality fails, even if one only considers the five types of modality enlisted by Helbig and Buscha (2001). Many more verbs can be found that express one of these modalities, notably brauchen ‘need’, vermögen ‘be able to’, haben + zu-infinitive ‘have to’, sein + zu-infinitive ‘is to’. Moreover, the concept of modality advocated by Helbig and Buscha (2001) is not systematic. As shown by Palmer (1986: 2), the notion of modality is rather vague and leaves a number of possible definitions open. First of all, it needs to be distinguished from the notion of mood. Whereas the term mood traditionally refers to an inflectional category, modality is typically marked by (modal) verbs, by particles and clitics Palmer (1986: 22). As Lyons (1977: 452) suggests, modality concerns the ‘opinion and the attitude’ of the speaker. In any case, modality covers much more phenomena than those taken into consideration by Helbig and Buscha (2001). A more systematic theory of modality would also have to consider verbal concepts as “try to”, “plan to”, “intend to”, “be inclined to”, “contemplate doing something”, “dare to” and many more. A corresponding concept of modality would concern an even larger number of verbs. A more elaborate but still rather extensional definition of modality is proposed by Portner (2009: 4), according to whom several subtypes have to be assumed that in turn involve a broad range of additional items to be considered: 1. sentential modality: modal auxiliaries, modal adverbs, generics, habituals, individual level predicates, tense and aspect, conditionals, covert modality 2. sub-sentential modality: modal adjectives, propositional attitude verbs, verbal mood, infinitives, dependent modals, negative polarity items

2.1 Traditional criteria |

3.

27

discourse modality: evidentiality, clause types, performativity of sentential modals, modality in discourse semantics

Finally, the major use of one of the six traditional modal verbs in Contemporary German is not captured by Helbig and Buscha’s conception: mögen in its prevailing use denotes affection. No matter what concept of modality one adopts, it would never constitute a homogeneous class that only comprises the six traditional modal verbs. Numerous approaches assume that the six traditional modal verbs differ from all of the remaining verbs in that these verbs, and only these verbs, express modality. The underlying concept of modality that these approaches rely on is an arbitrary enumeration of subtypes of modality. Accordingly, their concept of modality is not systematic. This holds true even for the most systematic attempt to establish a unified semantic analysis of the six traditional modal verbs, made by Bech (1949: 38). Being the first one who attributed the term Modalverb to the six verbs können, müssen, wollen, dürfen, sollen and mögen, he tries to collect all of the possible readings they occur with. In a second step he groups them into three subclasses: volitional (wollen, sollen, dürfen), emotives (mögen) and causal modals (können, müssen). They are further specified by means of two oppositions. The first one divides active modals (such as müssen) from passive ones (such as können). This roughly corresponds to the partition into necessity versus possibility modals, as proposed by Kratzer (1978) and Kratzer (1981). The second one determines whether the source of volition (or emotion) is located within the grammatical subject or subject external. Bech’s approach is inductive. He assumes a class of modal verbs consisting of six items. He then tries to extract all of the semantic properties they have in common. As illustrated above, the outcome is somehow biased. However, it remains unclear why Bech (1949) chose exactly these six verbs. It should not be surprising to see that Bech’s choice was arbitrary. As already indicated by Welke (1965: 19), a definition of a class of modal verbs with the traditional extension based on semantic grounds fails. The fact that authors often presuppose some concept of modal meaning without giving a clear definition, such as Fritz (1997: 13) and Diewald (1999: 299), deserves closer attention. It is not surprising that such a vaguely defined concept causes so much confusion. Johnen (2003: 11) reports that based on a similar semantic definition of about 230 different verbs are considered to be modal verbs in Portuguese, whereas two of them only carry auxiliary-like properties. Apart from the work by Kratzer (1978), Kratzer (1981) and Kratzer (1991), there is hardly any other attempt that tries to explicitly define modal verbs. Following the tradition of modal logic, she adopts a possible world semantics. More precisely, Kratzer (1981, 1991: 649) demonstrates that each modal verb can be iden-

28 | 2 Modal verbs: A class struggle

tified by means of three dimensions: (i) the modal force is typically either instantiated as universal quantification over possible worlds (necessity) or existential quantification over possible worlds (possibility), (ii) the modal base governs the composition of the set of worlds over which the modal verb quantifies: circumstantial modal verbs operate on worlds which describe the circumstances of the external world, epistemic modal verbs quantify over worlds which describe an epistemic state, (iii) the ordering source which introduces an ethical or volitional ideals according to which the set of possible worlds in the modal base are ordered. The main advantage of this theory is that elements that allow for different modal interpretations can be treated as uniform lexical entries which have a common and stable modal force, but which are specified for more than one modal base or ordering source. For this reason, Kratzer’s account will serve as the reference frame work for the study presented here. Since Kratzer’s approach implies a much broader concept of modality that applies to many more items than the traditional six modal verbs, she does not conform to the class of modal verbs in its traditional extension. Despite its popularity, Kratzer’s theory has remained incomplete since the early 1980s and has not undergone any substantial revisions. In strict contrast to Kratzer (1978, 1981, 1991), a new approach developed by Lassiter (2011) is based on the assumption that modal operators in general do not involve quantification over possible worlds. Based on data from modal comparatives (ϕ is at least as likely as ψ; it is better to trespass than it is to murder) and degree modification (ϕ it is 90% certain; I want very much to travel to Cuba), Lassiter (2011: 51–63,141–150) argues that modal operators denote measure functions on propositions. In his approach, modal operators introduce a scale propositions, which are ordered with respect to their probabilities (epistemic) or subjective or moral preferences which are weighted according to their probabilities. For instance, a necessity modal introduces a very high threshold value on this scale, which values all propositions as false that are not in the top most region of the scale. In essence, Lassiter’s work is a critique on Kratzer’s concept of the ordering source, which he consequently replaces by a couple of different measurement functions. Even if Lassiter’s analysis correctly points out a whole range of difficulties for Kratzer’s theory fragment with respect to modal comparison and degree modification, Lassiter’s approach makes problematic predictions regarding modal auxiliaries. Firstly, it has to be highlighted at this point that Lassiter’s analysis is primarily based on modal adjectives, rather than modal verbs. As Lassiter (2011: 89–93, 144) acknowledges, there is no empirical evidence that the English epistemic modal auxiliaries must, should, might and their deontic counterparts must and may de-

2.1 Traditional criteria |

29

note scales. Rather, the opposite is the case, as they neither participate in modal comparision, nor can they be the target of degree modifiers. Furthermore, Lassiter (2011: 132) notices that epistemic modal auxiliaries are upward monotonic, which he considers as a main characteristic for a quantifier. Yet, he concludes for mere theory internal reasons that auxiliaries have to denote scales rather than quantification over possible worlds, though he admits that a quantificational approach for modal auxiliaries in English is not totally excluded. Secondly, Lassiter (2011: iii-iv, 66–69, 154–164) assumes that epistemic and deontic modal operators involve substantially different types of scales and mechanisms of interpretation. Whereas the former denote ratio scales, the latter denote interval scales which are probability weighted. Accordingly, Lassiter (2011: 99) acknowledges that deontic modals and their epistemic counterparts operate on fairly different domains, and as a consequence, a uniform analysis for ambiguous modal verbs appears hard to maintain. This is an unwelcome side effect. As long as there is no compelling evidence that modal auxiliaries are interpreted relative to scales, a quantificational approach appears to be preferable for modal auxiliaries and verbs. 2.1.3.2 The expression of the possibility or necessity of the embedded predicate denotation Becker (1836: 176 §91–§93, 1841: 219) is one of the first grammarians who investigated auxiliary-like verbs in German that exhibit the IPP from a semantic perspective. As he observes, verbs like können, müssen, wollen, dürfen, sollen, mögen and also lassen have lost their lexical meaning and only denote an abstract semantic relation:²² aber sie drücken in ihrer jetzigen Bedeutung nicht mehr den Begriff eines Prädikates aus, sondern bezeichnen nur Beziehungsverhältnisse, nämlich die Möglichkeit und Nothwendigkeit der prädizirten Thätigkeit, die wir oben als Modusverhältnisse des Prädikates bezeichnet haben (§. 59) z.B. „Er kann tanzen” „Er muß husten”; sie werden daher Hülfsverben des Modus genannt.

The extent to which Becker’s (1836, 1841) approach anticipates the spirit of modern modal logic analyses, such as the one suggested by Kratzer (1978) and Kratzer (1981), deserves closer attention. The basic concepts are necessity and possibility.

22 “[...] but, in their contemporary usage, they no longer express the notion of a predicate, but merely denote relations, namely the possibility and the necessity of the predicated activity, which we called the mood of the predicate above (§ 59), e.g. “He can dance”, “He must cough”; they are therefore called auxiliaries of mood.” [own translation]

30 | 2 Modal verbs: A class struggle

Much like Kratzer’s modal base, Becker (1841: 221) suggests three types of specifications. Accordingly, necessity or possibility can be specified as real, moral or logical, whereas the last type corresponds to epistemic modality. Becker (1841) is one of the first grammarians who observes this type of modality. Becker (1841) is not only the first one who attempts to give a general semantic description of these seven items that is based on the concept of necessity and possibility, but also the first who adopts the term Modus ‘mood’ to denominate these seven auxiliary-like verbs. It is fairly likely that Becker (1841) is even the origin of the contemporary concept of the modal verb class. It is only a small step from his original term Huelfsverben des Modus ‘auxiliaries of mood’ to Modalverb, as it is used by Bech (1949). Even if Becker (1841) is on the right track, some amendments still have to be made. He proposes a very clear definition of modality in terms of necessity and possibility; yet, it remains mysterious how the volitional use wollen and the emotive use mögen fit into this picture. Moreover, his definition also applies to a whole range of other verbs. 2.1.3.3 The availability of an epistemic interpretation The most viable semantic criterion that can be invoked to justify the establishment of an independent class of modal verbs is the availability of an epistemic interpretation. It took a fairly long time in grammatical research until the peculiarity of these readings was acknowledged. Probably, the first one to consider epistemic readings as a general property of modal verbs is Becker (1841: 221), who briefly discusses the so-called logical possibility and necessity readings for kann, dürfte, muß, will, soll and mag. (23)

Er kann (dürfte, muß, soll) schon angekommen sein. he can might must shall already arrived-ppp be-inf ‘He could/might/must/is said to have already arrived’

(24)

Man will ihn gesehen haben. one wants him see-ppp have-inf ‘Somebody claims to have seen him’

In contrast to a moral or real possibility, kann in example (23) denotes a logical possibility. This latter type expresses that, in view of what he knows, the speaker considers it possible that the propositional content of the modified clause holds. As Becker (1836: 180) already argues, the logical modal verbs differ with respect to subtleties in their interpretation: kann refers to a possibility; dürfte to a probability; mögen always has a concessive resonance; muss refers to a logical necessity evaluated by the speaker; wollen expresses a logical necessity assessed by the

2.1 Traditional criteria |

31

subject referent and sollen expresses a logical necessity evaluated by another referent. For a couple of decades, until the beginning of the 20th century, epistemic readings did not attract too much attention. At best they are mentioned, but their particular status remains veiled. As one among few, Curme (1922: 319) enumerates the epistemic interpretation for each of the six traditional modal verbs, but he does not pay any further attention to them, just as Bech (1949) does not. Most grammars, such as Vernaleken (1861), however, go as far as to ignore the epistemic interpretation completely. Only in the early 1960s did Griesbach and Schulz (1960: 65) acknowledge the availability of an epistemic reading as an essential characteristic of the six traditional modal verbs. They are the first who systematically describe this type of modality for modal verbs in German. In their opinion, modal verbs are characterised by the availability of two different interpretations: an objective (non-epistemic, root) one, and a subjective (epistemic) one. Their position has frequently been adopted, e.g. by Öhlschläger (1989: 132), Engel (1996: 463), Diewald (1999: 1) and Reis (2001: 287). In more recent research this property has been referred to as polyfunctionality. In a less explicit way, Erb (2001: 74) also makes use of this concept. As Westmoreland (1998: 12) and Ziegeler (2006: 90) point out, epistemic modifiers are subject to a particular condition. Since they label the modified proposition as a mere assumption of the speaker, it follows that the epistemically modified proposition is not part of the speaker’s knowledge. Accordingly, whenever a speaker utters an epistemically modified proposition epistemic (p) such as the examples in example (23), he signals to the hearer that p is not part of his knowledge. It would cause quite some confusion if the speaker were to resume the discourse saying “. . . since I know that p is the case”. Canonically, the speaker would not know that p is false either. Likewise, he could not continue uttering “. . . although I know that p is not the case”, at least if he uses an epistemic modal verb which is inflected for the indicative. Similar observations have been made by Erb (2001: 161), Krämer (2005: 60, 133), Fintel and Gillies (2010: 353), Kratzer (2011, 2012: 99) and Martin (2011: Sect. 3.1), and a detailed discussion is given in Section 6. To a lesser extent, a similar position is defended by Papafragou (2006: 1693). In the remainder of this study, it will be demonstrated that the relevant referent does not always have to be the speaker, e.g. in embedded clauses or in information seeking questions. Accordingly, this condition will be formulated with respect to a more abstract expression. In his Lectures on Deixis in the early 1970s, Charles Fillmore introduced the concept of a deictic centre (in the reprinted version: Fillmore (1997: 98)), which was subsequently developed in more detail by Levinson (1983: 64). The deictic centre is a referent who is identical to the speaker in the most prototypical context, but it can be instantiated by a referent other than the speaker of the actual utterance. A similar concept had already been suggested

32 | 2 Modal verbs: A class struggle

by Bühler (1934: 102). According to his terminology, this referent is called Origio, and as Abraham (2011: xxxv) points out it can also be used to describe epistemic modality. As will be shown in the Chapter 4, it is much more appropriate to formulate the condition for epistemic operators with respect to deictic centres rather than with respect to the actual speaker. For the sake of simplicity, this condition will be referred to as ‘Condition on Deictic Centres (CoDeC)’ here: (25)

Condition on Deictic Centres (CoDeC) The use of an epistemic operator indicates that the embedded proposition is not part of the deictic centre’s knowledge.

Indeed, epistemicity turns out to be a property that does not apply to a lot of verbs in German. Apart from the traditional six modal verbs, only five more verbs come into consideration: brauchen ‘need’ (cf. Takahaši (1984: 21), Engel (1996), Askedal (1997a: 62)), werden (Vater (1975), Engel (1996), Enç (1996), Erb (2001: 176)), scheinen (Askedal (1998: 61), Wurmbrand (2001: 205)), drohen and versprechen (Askedal (1997b), Wurmbrand (2001: 205)). In some rare cases, even lassen exhibits an epistemic reading, as pointed out by Reis (2001: 308). It is a matter of debate to what extent these items really belong to the same class as the six traditional modal verbs. First of all, there is no agreement as to whether all of these items indeed involve epistemic semantics. Öhlschläger (1989: 8) denies that brauchen allows for an epistemic interpretation, and Reis (2005b) argues that drohen and versprechen should be considered as aspectual verbs rather than epistemic modal verbs. Secondly, some authors assume that the class defining property for modality is poly-functionality. Therefore, they reject all verbal items that do not involve both types of modality. According to Öhlschläger (1989: 8), brauchen cannot be regarded as a modal verb since it lacks an epistemic reading, whereas werden has to be excluded because of the absence of a circumstantial interpretation. Since the question which of these items indeed involve epistemic modality requires a thorough investigation of empirical data, it will be postponed until Section 2.2, where each verb will be individually reviewed with respect to the CoDeC. Even if the availability of an epistemic interpretation appears to be a powerful criterion, it does not apply equally to all of the six traditional items. Firstly, the canonical non-circumstantial uses of wollen ‘claim to’ and sollen ‘is said to’ differ from genuine epistemic modals. While the latter refer to a conclusion that is drawn by the speaker, the former express a claim by the subject referent (in the case of wollen) or some non-specified source (sollen). As Reis (2001: 294) points out, these instances of wollen and sollen are acceptable to a greater degree as non-finite forms than epistemic modals are. This might be due to the fact that they involve

2.1 Traditional criteria |

33

more argument structure than their epistemic counterparts: wollen is a control verb and has a subject argument of its own, and sollen contains some unspecified covert argument. Hence, there are plenty of reasons to treat these latter readings separately from epistemic modality. This type of approach is furthermore supported by the observation that they might violate the CoDeC. Because they are always related to some claim, they will be referred to as reportative in the remainder of this study. Secondly, dürfen can never be interpreted epistemically unless it is inflected for past subjunctive (dürfte). As will be shown in Section 2.2.5, deontic dürfen and epistemic dürfte differ with respect to the modal force they carry: deontic dürfen is a prototypical possibility modal verb, epistemic dürfte appears to express a stronger modal force than that. Therefore, dürfte should be considered as an independent lexical item. Thirdly, most contemporary researchers treat möchte as an independent lexical item as well, such as Öhlschläger (1989: 7), Kiss (1995: 162), Fritz (1997: 103), Diewald (1999: 144), Axel (2001: 40) and Wurmbrand (2001: 183). Since none of them provides evidence that it is used with an epistemic interpretation, strictly speaking, it cannot be considered as a modal verb. Regardless of these discrepancies, the availability of an epistemic interpretation turned out to be the most promising property to define a class. If the class of modal verbs is defined based upon epistemicity, only a small group of verbs comes under consideration. In the upcoming Section 2.2, all of these potentially epistemic verbs will be carefully reviewed with respect to the CoDeC. However, this approach will not result in a modal class with its traditional extension.

2.1.4 Conclusions As has been shown, the six traditional modal verbs do not form a class that can empirically be justified. All of the criteria that come into consideration fail. This includes morphological criteria (preterite present paradigm, obligatory IPP), syntactic criteria (sub-categorisation of bare infinitives), and semantic criteria (availability of an epistemic interpretation). Therefore, a number of authors have already conceded that the class of modal verbs in its traditional extension is arbitrary and not well defined, such as Welke (1965: 12), Birkmann (1987: 5), Öhlschläger (1989: 7) and Fritz (1997: 14). Thus, it becomes clear why different authors assume classes of modal verbs with diverging extensions. Some of these classes that have been a basis for influential theories are presented below: – Ehlich and Rehbein (1972: 318) modal verbs in German: müssen, können, dürfen, sollen, wollen, möchte, nicht brauchen, werden – without mögen

34 | 2 Modal verbs: A class struggle

– – – – –

Kratzer (1981: 40) modal auxiliaries in German: muss, kann, darf, soll, wird, mag, müßte, könnte, dürfte, sollte, würde, möchte – without wollen Kratzer (1991: 650) some modals: muss, kann, soll, wird, dürfte Fritz (1991: 46): epistemic modals in Contemporary German: dürfte, kann, könnte, mag, muss, müßte, soll, will, wird Wurmbrand (2001: 137) modal auxiliares in German: dürfen, dürfte, können, möchte, müssen, sollen, wollen – without mögen Erb (2001: 75) modal verbs in German: können, müssen, dürfen, sollen, wollen, mögen, werden

These authors are not always explicit as to why they exclude some of the verbs that are traditionally considered as modal verbs. Since the traditional class of modal verbs cannot be empirically justified, one could argue for a mere extensional definition. This would be plausible if the six relevant verbs invariably involved auxiliary-like properties across the periods of German. But as it turns out, during the Old High German period, each of these verbs was grammaticalised to a different extent. Birkmann (1987) and Diewald (1999) agree that sollen, wollen and mögen were already highly frequent as grammaticalised verbs with infinitive complements and modal semantics in Old High German. In contrast, the remaining traditional modals können and müssen can hardly be found in such an auxiliary-like use in this period, or not at all, in the case of dürfen. As illustrated by Birkmann (1987), dürfen only started to select infinitive complements during the late Middle High German period. Apart from that, he points out that there is one more grammaticalised preterite present in Old High German that occurs with modal meaning: eigan ‘have’. As Wurzel (1984) shows, it appears that most of the preterite presents in Old High German involve too much lexical content in order to be considered as modal auxiliaries. This seems to contradict the position advocated by Fritz (1997: 13), who claims that all of the six traditional modal verbs already exhibited modal semantics in Old High German. But as was already discussed in Section 2.1.3.1, many authors use a rather fuzzy concept of modality. The two diverging positions thus do not need to be a contradictory at all. This again demonstrates that the six traditional modal verbs did not become what they are as a chunk, but rather each verb had its own individual development, at its own pace. Meanwhile, some modals got lost (such as eigan) or are likely to get lost (such as mögen), but there are also new members in the group, such as möchten, which has already developed a full paradigm, at least in spoken language, as shown by Vater (2010). The process of grammaticalisation turns out to be even more complex. When focusing on an individual verb, it is not obvious that it acquired all features of

2.1 Traditional criteria |

35

auxiliary-hood at the same time. Even if sollen was already highly grammaticalised in Old High German, it is recorded with IPP only in the 16th century, more than hundred years after its first occurrence with a modal verb. As a consequence it appears that each property of auxiliary-hood develops independently, and the sequence of acquisition may differ from verb to verb. A different attempt to argue for a complex definition that relies on a variety of features has been suggested by Zifonun (1997: 1253). She argues that modal verbs are the class of verbs that (i) select bare infinitives, (ii) lack imperative morphology, (iii) have a fully developed paradigm of tense morphology, (iv) lack arguments of their own and (v) are evaluated with respect to a conversational background. But as she acknowledges herself, none of these criteria hold without exceptions. These observations lead to the conclusion that there is no reason at all to maintain the class of modal verbs in its traditional extension. But there is an alternative. Since the availability of an epistemic interpretation is restricted to a rather small group of verbs, it could serve to establish a homogeneous class. This type of approach has been taken by Engel (1996: 463) and Reis (2001: 312). As shown above, there are more verbs than the six traditional modal verbs that come into consideration for an epistemic interpretation, in particular brauchen and werden. As a consequence, they have to adapt the extension of their class of modal verbs accordingly. Both authors argue that there is a strong link between the selection of a bare infinitive complement and the availability of an epistemic reading, cf. Reis (2001: 308). However, there are some minor discrepancies in the accounts of Engel (1996) and Reis (2001). First of all, they identify reportative modality contributed by wollen and sollen with epistemic modality, and second, there is only scarce evidence that brauchen indeed involves an epistemic reading that is subject to the CoDeC, as will be demonstrated in Section 2.2.9. Nevertheless, the analysis suggested by Engel (1996) and Reis (2001) will be adopted here. In the next section, it will be empirically reviewed in great detail, in order to find out which verbs are captured by this class definition. Before concluding this section, one last plausible criterion for auxiliary-hood will briefly be addressed. As Grimm (1822: 851) argues, an essential property of auxiliaries is that they are more frequent compared to lexical main verbs. According to him, auxiliaries are ‘verba welche sehr häufig gebraucht werden und statt ihrer lebendigen bedeutung abstracte begriffe annehmen’ (‘verbs that are used very frequently and that involve abstract concepts rather than their original lexical meaning’). This can easily be tested by using a corpus. Based on the data collected by Ruoff (1981), the most frequent verbs in German as spoken in Baden Württemberg (southwestern Germany) are the ones illustrated in Table 2.3. Once more, it turns out that the six traditional modal verbs behave

36 | 2 Modal verbs: A class struggle

Tab. 2.3: Most frequent verbs in spoken German, according to Ruoff (1981), based on a corpus study comprising 500,000 word form tokens sein (‘be’, prf.aux) haben (‘have’, prf.aux) kommen (‘come’) gehen (‘go’) müssen (‘must’) werden (‘become’, pass.aux, fut.aux) machen (‘make’) sagen (‘say’) können (‘can’) wissen (‘know’) tun (‘do’) geben (‘give’) wollen (‘want’) schaffen (‘manage’) kriegen (‘get’) fahren (‘drive’) heißen (‘mean’) nehmen (‘take’)

24.11 % 22.72 % 3.78 % 3.31 % 3.24 % 2.67 % 2.58 % 2.26 % 2.01 % 1.21 % 1.19 % 1.1 % 0.83 % 0.82 % 0.78 % 0.72 % 0.61 % 0.58 %

lassen (‘let’) stehen (‘stand’) sehen (‘see’) laufen (‘run’) sollen (‘shall’) bringen (‘bring’) kaufen (‘buy’) brauchen (‘need’) ziehen (‘draw’) dürfen (‘may’) glauben (‘believe’) helfen (‘help’) meinen (‘think’) ... möchten (‘want’) ... mögen (‘like’) ...

0.53 % 0.53 % 0.51 % 0.50 % 0.48 % 0.41 % 0.38 % 0.37 % 0.36 % 0.36 % 0.34 % 0.33 % 0.30 % 0.08 % 0.04 %

fairly differently. Whereas müssen and können are among the most frequent verbs, mögen is rather rare. As a consequence, frequency cannot serve to establish a class of modal verbs, either. This data is not reliable for a couple of reasons. Firstly, the corpus is too small to allow for any significant results, as it only comprises 500,000 word form tokens. Secondly, the annotation is not precise enough. As will be shown in Section 2.2, some of the verbs, such as können, mögen and wollen, also involve transitive uses or finite dass-clauses. In these instances the lexemes mentioned above clearly behave like main verbs, and accordingly, they cannot be regarded as auxiliaries. Therefore, these occurrences should be ignored. But nevertheless, even if only modal verbs with infinitives are considered, they are not more frequent than other common main verbs. Finally, some of the verbs listed here are part of lexicalised idiomatic expressions, such as heißen, which frequently occurs in das heißt ‘that is to say’. It is not obvious whether this can still be considered an occurrence of a main verb, or whether a reanalysis of the whole chunk to another category, such as a discourse marker, has taken place. Altogether, this small survey might appear imprecise, but at least it gives us an idea of the different frequencies of the individual modal verbs. Nevertheless, it turns out that a remarkably high percentage of the most frequent verbs involves predicate complex formation. Maybe it is possible to recast Grimm’s (1822) original

2.2 Case studies |

37

Tab. 2.4: Frequency of auxiliaries among all word form tokens, according to Kaeding (1897), based on a corpus study comprising 10,910,777 word form tokens sein (‘be’, prf.aux) haben (‘have’, prf.aux) werden (‘become’, pass.aux, fut.aux) können (‘can’) lassen (‘let’)

2.83 % 1.21 % 1.22 % 0.48 % 0.29 %

müssen (‘must’) sollen (‘shall’) wollen (‘want’) mögen (‘like’) dürfen (‘may’)

0.28 % 0.22 % 0.16 % 0.13 % 0.09 %

idea in new terms: Whenever a verb is frequently used, it is likely to undergo predicate complex formation, in the spirit of Höhle (1978), Haider (1993) and Haider (2010), Kiss (1995) and Müller (2002). Welke (1965: 19) refers to another corpus-based study on the frequency of modal auxiliaries that was carried out by Kaeding (1897). In his corpus that obviously consists of written texts, Kaeding extracts the following frequencies: können 52,384; lassen 32,143; müssen 30,350; wollen 27,834; sollen 23,910; mögen 14,406; dürfen 9 432. This tendency is reflected in the results of the small corpus study carried out by Diewald (1999: 9 ). She investigated a corpus that contained 839 modal verbs. The frequency of the different verbs is as follows: können 319 (38,02 %), müssen 182 (21, 69 %), wollen 152 (18,12 %), sollen 100 (11,92 %), mögen 48 (5,72 %) and dürfen 38 (4,53 %). Overall, similar pictures emerge: können and müssen are the most frequently used modal verbs, mögen and dürfen are used less frequently. However, there are some minor differences. There are a couple of ways to account for them. First of all, the corpora are composed of completely different registers: While Kaeding collected written language, Ruoff focused on spoken language of a variety spoken in southwestern Germany. Moreover, the data compiled by Kaeding is at least 100 years older than Ruoff’s data. This might already explain why the frequence of mögen is much higher in Kaeding’s corpus, since it was one of the dominant verbs in the earlier stages of German.

2.2 Case studies All of the different verbs discussed so far come in different guises. In the present section, the syntactic distribution of these verbs will be systematically reviewed. The following patterns will be taken into consideration: Transitive verbs, verbs with directional phrases, control verbs and raising verbs. As will be shown in Chapter 3, circumstantial modality can be rephrased as event modification, and

38 | 2 Modal verbs: A class struggle

epistemic modality as propositional modification. Authors who follow the tradition of Lyons (1977: 799) additionally distinguish between a ‘subjective’ and an ‘objective’ epistemic interpretation. However, as will be shown in Section 4.22, this distinction is misleading, and the assumption of a separate category ‘objective’ epistemic modality introduces more problems than it solves. Moreover, all of the different patterns under investigation will be illustrated by means of empirical data taken from the German reference corpus (DeReKo). It was created at the Institut für Deutsche Sprache Mannheim. At the time the study presented here was carried out, it comprised about two billion of word form tokens. In order to illustrate the nature of the verbs discussed below, it is sometimes important to take a diachronic perspective. Accordingly, historical data will be discussed at places. Occasionally, a brief view of the diachronic development of these items may reveal deeper insights on their nature in Contemporary German.

2.2.1 können Depending on its distribution, können contributes a couple of rather different semantic concepts, ranging from ability to epistemic possibility. In more syntactic terms, it can be realised as a transitive verb, as a control verb, and as a raising verb. In addition, it also occurs with verbless directional phrases. There is one particular use of können that reveals its underlying semantic nature: The quantificational use as discussed by Carlson (1977: 119) and Brennan (1993) and Brennan (1997). As Brennan (1993: 102) demonstrates, there is no plausible alternative to analyse this use of können as an existential quantifier (∃) that binds the variable contributed by the indefinite NP. This will also be demonstrated below. 2.2.1.1 Transitive uses There has been a great deal of discussion about the precise status of modal verbs without infinitival complements, as the instance of können in example (26) illustrates. (26)

Die Bewerber können Russisch. the applicants can Russian ‘The applicants can speak Russian’

As Öhlschläger (1989: 68) discusses, the essential question is how the absence of an infinitive can be accounted for. Two conflicting approaches have been taken so far. On the one hand, these occurrences could be treated as ellipsis of the infinitive, and as a consequence, the remaining objects would be arguments of the elided verb. This approach has been suggested by Heyse (1822: 403), who argues

2.2 Case studies |

39

that the traditional modal verbs always require an infinitive complement. On the other hand, one could assume that these cases are instances of transitive verbs. Accordingly, the remnant objects would belong to the (modal) verb. It is also important to ask what criteria can be used in order to understand the nature of these patterns. According to Öhlschläger (1989: 69), the essential question is whether the elided infinitive can be unambiguously determined. In the case of wollen without infinitive, it is always possible to insert the verb haben ‘have’ or bekommen ‘get’ in the gap. Therefore, Öhlschläger (1989) concludes that these cases of wollen involve ellipsis. As Öhlschläger (1989: 71) further argues, it is not so easy to reconstruct the elided infinitive in the case of können. Accordingly, he assumes that these instances are genuine transitive verbs. If any of the cases discussed above indeed involve ellipsis, (i) it should be possible to specify its particular type. Since canonical cases of ellipsis affect rather syntactic configurations than specific predicates, it is expected (ii) that all types of embedded infinitives (ditransitive verbs, verbs with dative arguments, verbs with genitive arguments, etc.) are affected to the same extent, and (iii) it should have no impact on the grammaticality of the entire utterance whether the matrix predicate is passivised or not. First, it appears that Öhlschläger (1989) uses a rather intuitive concept of ellipsis. In their typologies, Winkler (2006) and Merchant (2009) distinguish between six types of ellipsis. In contrast to the cases discussed here, all of their subtypes concern particular syntactic configurations that involve clausal conjunction. In particular, the elided phrase has an overt antecedent in the first conjunct. Moreover, there is no type that only affects an embedded infinitive without its complement. In gapping, the finite verb is elided (cf. 27) and, in VP-ellipsis, the entire VP (cf. 28; both examples taken from Winkler (2006)). (27)

Manny [plays]i the piano and Anna _i the flute.

(28)

They [play the piano]i but Anna doesn’t _i .

Modal verbs without an infinitive complement such as in (26) cannot be derived from VP-ellipsis in German for one additional reason. In contrast to English, the VP is phonetically not completely elided but realised as the VP-anaphora es, as López and Winkler (2000: 624) have pointed out. Thus, it becomes evident that the type of ellipsis that Öhlschläger (1989) has in mind would be completely different in nature. In a similar manner, können with an accusative NP cannot be analysed as a fragment in correspondance with the analysis suggested by Merchant (2009), as fragments are typically incomplete answers to wh-questions. Yet, it remains to be checked whether this pattern could be regarded as “situational ellipsis”, which is discussed by Schwabe (1994: 2).

40 | 2 Modal verbs: A class struggle

Secondly, this type of ellipsis would look suspicious because it only affects a semantically restricted group of embedded infinitives. If the matrix predicate is wollen, the elided infinitive could only be identified with haben ‘have’ or bekommen ‘get’. In the case of mögen, there is only one verb that comes into consideration: leiden ‘suffer/be seriously affected’. können turns out to be the least restrictive matrix verb; the gap could be interpreted as ellipsis of verstehen ‘understand’, sprechen ‘speak’ or machen ‘make’ (and related concepts). It should be noted that all of these verbs that can potentially be subject to ellipsis are transitive verbs that contribute precisely a NP with accusative case. This is somewhat unexpected. Indeed, it should be possible to elide any type of infinitive, irrespective of its argument structure. Interestingly, ellipsis does not apply as soon as the transitive verb is replaced with a non-transitive synonym, as is illustrated by the contrast between examples (29) and (30). Likewise, an infinitive needs to be realised whenever the sole argument is a dative NP, like in the case of helfen ‘help’ (cf. 31). (29)

Der Herbert kann Russisch (sprechen). the Herbert can Russian speak-inf ‘Herbert can speak Russian’

(30)

Der Herbert kann auf Russisch sogar über the Herbert can in Russian even about Wissenschaftsgeschichte #(sprechen). history.of.science speak-inf ‘Herbert is able to even talk about history of science in Russian’

(31)

Der Herbert kann dem David #(helfen). the Herbert can the-dat David help-inf ‘Herbert is able to help David.’

This behaviour is completely unexpected under an analysis which assumes ellipsis of the infinitive complement, even for one that assumes lexical licensing. In a similar vein, this type of approach could not explain why können in example (29) can only refer to an ability, but never to a permission, a practical possibility or an epistemic possibility. The latter three readings are always possible if the können occurs with an infinitive complement. Finally, it is not clear why ellipsis can never apply to two-place predicates that select a dative predicate such as helfen ‘help’. A further strong argument in favour of a transitive analysis comes from data about passivisation. In contrast to Welke (1965: 14) and Helbig and Buscha (2001: 116), who assume that modal verbs are generally incompatible with passivisation, instances of passivised forms of können can be found in corpora in which they behave like a prototypical transitive verb, such as in examples (32) or (33):

2.2 Case studies |

(32)

41

Auch dem Skispringer geht im Ernstfall eines even the ski.jumper goes in.the case.of.emergency a-gen Wettbewerbs ein Gutteil von dem verloren, was erlernt, competition-gen a bulk of that lost what learned, eingeübt ist – scheinbar so schlafwandlerisch gekonnt trained is apparently so somnambulisticly can-prt.pas wurde²³ pas.aux ‘Even the ski jumper forgets most of what he has learnt, what he did in training and what he had apparently been able to do in his sleep.’ lit: what he apparently could do in his sleep

(33)

So kann beispielsweise im Handel gepunktet werden, wenn so can for.example in.the commerce scored pas.aux if eine mitteleuropäische Sprache, eventuell auch Russisch, a central-European language maybe also Russian gekonnt wird²⁴ can-prt.pas pas.aux ‘For example, it can be advantageous in business if you can speak a central European language and perhaps even Russian’ lit: If you can speak a central European language

Defending an analysis that assumes ellipsis, one could argue that the infinitive has just been elided in these examples above. But, then, it is expected that cases of passives in which the infinitive is spelled out should be equally acceptable. As is shown in examples (4) and (35), this is clearly not the case:²⁵

23 DeReKo: N97/DEZ.51590 Salzburger Nachrichten, 22/12/1997. 24 DeReKo: P95/APR.14638 Die Presse, 28/04/1995. 25 Some authors, such as Zifonun (1997: 1255), Erb (2001: 90) Wurmbrand (2001: 202), argue that passives with overt infinitives are indeed acceptable: (1)

Auch sterben muß gekonnt sein/werden. also die-inf must can-prt.pas be-inf/pas.aux.pst.inf ‘You must learn the skill of death.’

(2)

? Karriere machen wird von Frauen oft gewollt. career make-inf pas.aux.pst by women often want-prt.pas

(3)

? Aufsätze schreiben wird heutzutage von keinem Schüler mehr essays write-inf pas.aux.pst nowadays by no pupil anymore gekonnt. can-prt.pas

‘Often Women want to make a career for themselves.’

42 | 2 Modal verbs: A class struggle

(34)

tun gekonnt. that pas.aux.pst do-inf can-prt.pas

* Das wurde

Intended interpretation: ‘They were able to do it.’

(35)

sprechen gekonnt. Russian pas.aux.pre speak-inf can-prt.pas

* Russisch wird

Intended interpretation: ‘They can speak Russian.’

Arguably, examples of passivised können without an infinitive complement are rare in written speech. This may be due to the fact that stative predicates such as verbs which encode a possession, knowledge or ability are not very compatible with passives in semantic respect, as the corpus data for English gathered by Gries and Stefanowitsch (2004: 108–110) show. One could therefore conclude that they are part of a phenomenon that is not relevant to grammatical description. How-

‘Pupils can no longer write essays in these days.’ However, in both cases, the grammaticality of the utterance is rather doubtful. Moreover, it is not clear whether the verbs sterben, machen and schreiben are indeed true infinitives rather than nominalisations. In all of these examples, the presence of an correspondingly inflected article das would considerably increase the acceptability of the sentence. Aside from that, all of these examples involve topicalisation of the infinitive complement and in the latter two cases, it is arguably a topicalisation of the entire VP. This also favours of an analysis that treats the topicalised infinitives as nominalisations that act as subject NPs. If (ii) and (iii) did indeed involve genuine infinitive complements, it is expected that the infinitive should be able to remain at the right periphery. In such a configuration, long passive should apply due to predicate complex formation and accordingly the NP Aufsätze would become the subject of the passive auxiliary werden and enter an agreement relation. (4)

* Aufsätze werden heutzutage von keinem Schüler mehr schreiben essays pas.aux.pst nowadays by no pupil anymore write-inf gekonnt. can-prt.pas ‘Pupils can no longer write essays in these days.’

Furthermore, the option with the copula sein in Zifonun’s (1997) example is clearly preferred. This is not surprising, since gekonnt in the clause above could also be interpreted as adjective. If so the whole pattern with the copula sein could be interpreted as stative passive which is syntactically something completely different from the canonical werden-passive, as shown in Maienborn (2007). The option with werden is at best acceptable if sterben is interpreted as nominalisation and if a corresponding article is introduced. All of these observations indicate that the instances of können and wollen in the passivisations above involve transitive patterns that select nominalised subject NPs. This is further supported by the fact that those of the traditional modal verbs that cannot be used in a transitive way, such as müssen would be even less grammatical in the environments above.

2.2 Case studies |

43

ever, the reason why they do not occur so frequently might be pragmatic. Besides können, there are typical transitive verbs that cannot often be found in passivised forms such as the preterite wissen, as has been pointed out by Adelung (1801: 1581). Nevertheless, passivised examples of können exist. And as opposed to other members of the traditional modal verb class, they prove much more acceptable. Similar contrasts have been shown by Hetland and Vater (2008: 104). (36)

* Russisch wird

(von allen) gemusst Russian pas.aux.pre by everyone must-prt.pas

‘Everyone must speak Russian.’

(37)

* Russisch wird

(von allen) gesollt Russian pas.aux.pre by everyone shall-prt.pas

‘Everyone shall speak Russian.’

(38)

* Russisch wird

(von allen) gedurft Russian pas.aux.pre by everyone may-prt.pas

‘Everyone is allowed to speak Russian.’

In order to account for all of these contrasts, it is necessary to accept that there are transitive uses of können. This is further supported by the diachronic development of the traditional six modal verbs. As observed by Fritz (1997: 68), the predecessors of the traditional six modal verbs in Contemporary German occur occasionally with NP complements in Gothic. Birkmann (1987: 118) observes that ¯ the Gothic θaurban ‘need’, kunnan, gamotan ‘have freedom’ and the Old High German kunnan and thurfan occur as lexical main verbs (Birkmann (1987: 161)). In a similar vein, Abraham (2004: 137) observes that in general non-finite complements have developed out of nouns. Whereas zu-infinitives have their origin in noun phrases marked with the dative case, bare infinitives emerged out of noun phrases with the accusative case. This is further enforced by Paul (1920: 95), who demonstrated that the bare infinitive complements originally go back to an NPcomplement with accusative. Here are some examples for the transitive kunnen from Middle High German from the late 12th century: (39)

kan.²⁶ sprâche oder swer hie welsche or whoever here foreign-acc language-acc can ‘or whoever here can speak a foreign language’

(40)

ine kan decheinen buochstap.²⁷ I.neg-cl can no-acc letter-acc ‘I don’t know any letter - I am illiterate.’

26 Wolfram, pârzival, 115,27 (1200). 27 Wolfram, pârzival, 115,27 (1200).

44 | 2 Modal verbs: A class struggle

(41)

leider des enkan er niht.²⁸ unfortunately this-gen he neg-cl.can neg ‘Unfortunately, he is not able to do this.’

(42)

ruochts got, ich pin vor valsche vrî: // ich enkan wants.it God I am for deceitfulness free I neg.cl-can decheinen widersaz.²⁹ no-acc hostility-acc ‘God willing, I am free of any deceitfulness, I am not able to commit any hostility.’

(43)

got noch künste kan genuoc.³⁰ God still arts-acc can enough ‘God is full of powers.’

As is shown here, kunnen was originally a transitive verb and acquired its infinitive complement only in the course of time. This observation is further supported by Diewald (1999: 34), who assumes that the modal verbs in Contemporary German generally started out as lexical main verbs that belonged to the class of transitive verbs. In some rare cases, they occur even with passive morphology. This leads to the conclusion that these instances have to be analysed as true transitive verbs instead of an ellipsis of the infinitive. An approach in terms of ellipsis could only be maintained under the assumption that the ellipsis is licensed by the lexical semantic of the matrix verb. It would then be necessary to add to the respective lexicon entries the precise information under which conditions an ellipsis is licensed. In particular, this concerns the infinitives that can be elided under the respective matrix verb. It might turn out that this type of approach is fairly laborious. Unless there is no concrete attempt to investigate the precise conditions of licensing, it is safer to assume that all of these examples of the use of können without an infinitive are transitive uses, as has been assumed by Öhlschläger (1989: 69) and Erb (2001: 96). Finally, Becker (1836: 178) has already observed that the transitive use of können appears to have a very specific meaning: It always seems to express knowledge rather than a physical ability. This is on par with its etymology: originally, it referred to a mental knowledge rather than to a physical ability. For the latter purpose, its counterpart mögen was used.

28 Wolfram, pârzival, 193,09 (1200). 29 Wolfram, pârzival, 439,21 (1200). 30 Wolfram, pârzival, 796,16 (1200).

2.2 Case studies |

45

2.2.1.2 Control infinitives with event modification The essence of being a control verb is to contribute an independent subject argument. In the case of the traditional modal verbs, this subject argument is also the source of modality. According to Palmer (1986: 16) based on Wright, these cases are the proto-typical instance of dynamic modality. In its dynamic interpretation, können ascribes an ability to the matrix subject. Therefore, it needs to carry an independent subject argument to identify the ability-holder. (44)

„Ich kann mir den Höhenflug nicht erklären” I can refl the altitude.flight neg explain-inf zu Guttenberg über seinen Wirtschaftsminister minister.for.economic.affairs zu Guttenberg about his Aufstieg.³¹ advancement ‘ “I can’t explain my success” Minister for Economic Affairs zu Guttenberg about his advancement.’

(45)

Ihre Bedeutung ist allerdings so sehr vom Zusammenspiel its meaning is however so much from.the interaction mit anderen Faktoren abhängig, dass nur professionelle with other factors dependent that only professional Penisleser sie richtig auslegen können.³² penis.readers they correctly interpret-inf can ‘However, its meaning is related to so many other factors that only professional penis readers can correctly interpret them.’

There are at least three ways to determine whether a verb is a control predicate or not. Control predicates (i) should not embed infinitives that lack a referential subject, (ii) they should not tolerate de dicto interpretations of quantified subjects and (iii) they should not exhibit voice transparency. Note, however, that all of the verbs under investigation here are ambiguous between numerous interpretations that often differ only in subtleties. In the case of können, there are at least three readings that partly overlap: The ability of the subject versus general possibility versus epistemic possibility. Whenever können denotes an ability of its subject referent, it cannot embed predicates without referential subject arguments (cf. 46) or predicates without subjects (cf. 47). Accordingly, if the raising test employed by Perlmutter (1970: 108), Brennan (1993: 41), Pollard and Sag (1994: 137), Axel (2001: 39), Wurmbrand (2001: 189), Erb (2001: 85), Müller (2002: 46, 2007: 256) cannot be ap-

31 DeReKo: NUN09/SEP.01543 Nürnberger Nachrichten, 14/09/2009. 32 http://astrogenital.de/html/penislesen.html, accessed in November 2010.

46 | 2 Modal verbs: A class struggle

plied, which indicates that this use of können is a control verb. Nevertheless, the general possibility reading remains available. (46)

# Es kannABIL schneien.

it can

snow-inf

‘Intended reading: It has the ability to snow.’

(47)

# ... dass getanzt

werden kannABIL . that getanzt-prt.pas pas.aux.inf can.

‘Intended reading: ...that it has the ability to be danced’

Of course, whenever there is no referential subject argument present, the dynamic use of können fails to identify the bearer of the ability. Moreover, Stechow (2003: 203) points out that control verbs that take quantified subjects only allow de re interpretations in which the quantified subject takes scope over the modal operator. This is expected, since the control verb has a subject argument of its own and, as a consequence, this argument can never be interpreted as the subject of the embedded infinitive. In contrast, de dicto interpretations, in which the modal verb takes scope over the quantifier in subject position, are excluded. This is only possible if the quantified subject originally belongs to the embedded infinitive – a configuration that is excluded with control verbs: (48)

Kein Student kannABIL Dänisch (sprechen). no student can Danish speak-inf OK: de re, ¬ > MV, ‘For no studenti : xi has the ability to speak Danish.’ #: de dicto MV > ¬ ’xi has the ability that ’

Finally, as has been demonstrated, among others, by Newmeyer (1970: 191), Jackendoff (1972: 105), Ebert (1976: 39), Öhlschläger (1989: 77), Pollard and Sag (1994: 136), Kiss (1995: 163), Diewald (1999: 62), Erb (2001: 92), Reis (2001: 301, 2005: 139), Stechow (2003: 205) and Hornstein (2003: 8), control verbs are not transparent with respect to voice. A sentence which contains an active verb (cf. 49a) always expresses the same proposition as its the corresponding sentence which contains the corresponding passivised verb (cf. 49b). Whenever they are embedded by a control predicate, the subject of the clause is assigned an additional semantic role, the one that is contributed by the control predicate. Note that active and passivised clauses differ with respect to the argument that is realised as the subject. In each case a different argument will be semantically marked as the subject argument. Therefore, control predicates are not transparent with respect to voice.

2.2 Case studies |

(49)

a.

47

Der Reinhold bezwingt den Nanga Parbat ohne the Reinhold conquers the-acc Nanga Parbat without Sauerstoffgerät. oxygen.apparatus ‘Reinhold conquers the Nanga Parbat without oxygen apparatus.’

b.

Der Nanga Parbat wird vom Reinhold ohne the-nom Nanga Parbat pass.aux by.the Reinhold without Sauerstoffgerät bezwungen. oxygen.apparatus conquer-ppp ’The Nanga Parbat is conquered by Reinhold without oxygen apparatus.’

(50)

a.

Der Reinhold kannABIL den Nanga Parbat ohne the Reinhold can the-acc Nanga Parbat without Sauerstoffgerät bezwingen. oxygen.apparatus conquer-inf ‘Reinhold has the ability to conquer the Nanga Parbat without oxygen apparatus.’

b.

# Der

Nanga Parbat kannABIL vom Reinhold ohne the-nom Nanga Parbat can by.the Reinhold without Sauerstoffgerät bezwungen werden. oxygen.apparatus conquer-ppp pass.aux-inf

’Intended reading: The Nanga Parbat has the ability to be conquered by Reinhold without oxygen apparatus.’

Returning to the ability reading of können, the source of the ability is identified with Reinhold in example (50a). In contrast, the source of ability would be assigned to the Nanga Parbat in example (50b). Since it is not clear whether mountains can be regarded as legitimate bearers of abilities, the interpretation of example (50b) is rather odd. A similar discussion is summarised by Reis (2001: 301). In essence, the two sentences are not semantically equivalent; accordingly, könnenABIL has to be considered as a control verb. In a similar vein, Wurmbrand (1999: 604, 2001: 199) argue that control verbs generally fail to embed passivised complements. Brennan (1993: 45) applies a similar test that employs symmetric predicates instead of passivisation. There is a widespread consensus to analyse the ability reading of können as a control structure, as has been suggested by Welke (1965: 49), Stechow and Sternefeld (1988: 429), Wurmbrand (2001: 171), Axel (2001: 40), Reis (2001: 302) and Erb (2001: 78).

48 | 2 Modal verbs: A class struggle

2.2.1.3 Raising infinitives with event modification There has been a great debate on the syntactic nature of non-epistemic (circumstantial or root) modal verbs. As Reis (2001: 300) demonstrates, three main positions can be distinguished. Originally, all circumstantial modal verbs were considered to be control verbs. In contrast to that, epistemic modal verbs were treated as raising verbs. The basic idea dates back to Ross (1969: 86), who assumes that root and epistemic modal verbs are derived from different deep structures. Whereas root modals originate from two place predicates, epistemic modals involve one place predicates. This idea was further developed by Jackendoff (1972: 102), Brennan (1993: 25), Stechow and Sternefeld (1988: 429), Diewald (1999: 62) and, finally, Abraham (2001: 18; 2002: 36; 2005: 241, 257, 261). Often, these approaches have not been systematically developed and remain rather superficial. In particular, this concerns Jackendoff (1972: 102) and Stechow and Sternefeld (1988: 429), who discuss only one or two items, and who fail to give an exhaustive description of the entire group of the traditional modal verbs. Most of these authors acknowledge that circumstantial modals can also be found in raising configurations. Following the observations discussed by Newmeyer (1970: 191), Jackendoff (1972: 105) was already aware that, with respect to voice transparency, modal verbs behave like raising verbs whenever the object of the embedded infinitive is inanimate, such as the Nanga Parbat in example (50). As noticed by Stechow and Sternefeld (1988: 446), each modal verb can also be realised as a raising pattern. Finally, Brennan (1993: 27) concedes that ought-tobe-deontics have to be analysed as raising verbs, following the assumption made by Feldman (1986: 177), who argues that they must be one-place predicates. This type of approach faces additional difficulties since the use of (reportative) wollen, which is traditionally considered an epistemic modal verb, involves a control pattern. This will be thoroughly discussed in Section 2.2.3. A second tradition, represented by by Welke (1965: 49) and Höhle (1978: 81, 84), argues that most circumstantial modals are even one place predicates without a subject argument of their own. This approach has been adopted and developed by Öhlschläger (1989: 105), Palmer (1990: 47), Geilfuß (1992), Kiss (1995: 163), Axel (2001: 40), Reis (2001) and Erb (2001: 73). According to these authors, some circumstantial modals such as volitional wollen and möchte are nevertheless to be analysed as control patterns. The third position defended by Wurmbrand (1999, 2001: 201) and Gergel and Hartmann (2009: 327) assumes that modal verbs are always raising verbs. Whereas Wurmbrand (2001: 201) analyses volitional verbs like wollen and möchte rather as main verbs, Gergel and Hartmann (2009) assume that they involve oblique raising into theta positions. Finally, Barbiers (2002: 67)

2.2 Case studies |

49

argues for an intermediate position claiming that all circumstantial modals occur with both patterns: control and raising. Returning to können, it appears that its use as a raising verb covers a broad range of different readings. Notably, it denotes a practical possibility. Moreover, it can involve deontic permission readings and quantificational readings. Applying the same diagnostics for the existence of a subject argument that were discussed in the last section, it turns out that these instances of können behave fairly differently. First of all, they do not require referential subjects any longer and as a consequence the tests proposed in the preceding section can be applied with out any problem. Such instances can also be found in corpora, as is indicated in examples (55)–(56). (51)

Es kannPERM (hier ruhig) schneien. it can here par snow-inf ‘It’s okay if it snows here.’

(52)

... dass (ruhig) getanzt werden kannPERM . that par getanzt-ppp pass.aux-inf can. ‘It’s okay if you dance here’

(53)

Es kannPOSS (hier manchmal) schneien. it can here par snow-inf ‘It can snow here from time to time.’

(54)

... dass (hier manchmal auch) getanzt werden kannPOSS . that dance-ppp pass.aux-inf can. ‘People dance here from time to time.’

(55)

Reisezeit: Der Montblanc lässt sich am besten von Ende travel.season the Mountblanc let relf at.the best from end Juni bis Anfang September umrunden. Auch im Hochsommer June until begin September round also in.the midsummer kann es schneien.³³ can it snow-inf ‘Travel season: for those who want to hike around the Montblanc, it is recommended to envisage this tour between the end of June and the begin of September. But it can also snow in midsummer.’

33 DeReKo: RHZ11/AUG.09341 Rhein-Zeitung, 09.08.2011.

50 | 2 Modal verbs: A class struggle

(56)

Zwar sprechen viele Holländer gut Deutsch, dennoch kann es indeed speak many Dutch well German yet can it Mißverständnisse geben, wenn Gleiches unterschiedlich misunderstandings give-inf if same differently gedeutet wird.³⁴ interpreted is ‘Many Dutch may speak German very well, but there can be misunderstandings if the same things are interpreted in a different manner.’

(57)

In diesem Raum kann gewohnt, gefeiert oder geschlafen in this room can live-ppp celebrate-ppp or slept-ppp werden.³⁵ pass.aux-inf ‘In this room, it is possible to live, to celebrate or to sleep.’

In its permission reading, können has a meaning similar to dürfen. The presence of the modal particle ruhig in examples (108)–(109) indicates that the interpretation of the possibility verb können has to be a deontic (permissive) one, as has been demonstrated by Grosz (2014: Sect. 5.1): As soon as ruhig is used in an utterance a modal possibility operator cannot be interpreted in an epistemic way. The possibility reading of können can be paraphrased in terms of temporal quantification, such as ‘From time to time, it happens that...’, this interpretation is highlighted by the presence of the adverb manchmal ‘occasionally’ in examples (53)–(54). As soon as können carries no subject arguments of its own, de dicto readings become available, as has been illustrated by Wurmbrand (1999: 606, 2001: 192) and Stechow (2003: 203). This type of interpretation obviously needs to be accompanied by a corresponding intonation contour, in which the negative subject quantifier is set off by a small break and receives a high pitched accent. Likewise, Blühdorn (2012: Sect. 8.5/364) has pointed out that a narrow scope interpretation of the negation becomes more likely once it bears a high pitch accent (H*L). (58)

Es kann (auch) | KEINH*L Student kommen. expl can (also) no student come-inf OK: de re, ¬ > MV, ‘For no studenti it is allowed/possible that hei comes.’ OK: de dicto MV > ¬ ’it is allowed/possible that no student comes ’

De dicto readings with the deontic pattern können are also possible with existential quantifiers. In contrast to examples (58), such configurations can easily be found in corpora. In example (59), the deontic possibility operator takes scope

34 DeReKo: R98/SEP.75404 Frankfurter Rundschau, 19.09.1998. 35 DeReKo: M10/APR.26150 Mannheimer Morgen, 06.04.2010.

2.2 Case studies |

51

over the existential quantifier ein Elternteil (□ > ∃). The possibility to access the benefit is granted for one of the two parents and it is not specified whether it has to be the mother or the father.

Die staatliche Leistung wird maximal 14 Monate gezahlt. Ein the public benefit is maximally 14 months paid a Elternteil kann sie aber längstens ein Jahr in Anspruch nehmen. parent can she but at.most a year in claim take-inf Die weiteren zwei Monate gibt es nur, wenn dann der Partner das the further two month gives it only if then the partner the Kind betreut.³⁶ child cares

(59)

‘The public benefit will be paid for maximally 14 months. A parent can draw it for no longer than a year. The extra two months are only available, if the other partner takes care for the child during that period.’

Finally, the permission reading and the possibility reading are both transparent for voice. Both examples (60) and (61) turn out to be synonymous. In order to unambiguously trigger the respective reading, more plausible contexts have been chosen. However, as Reis (2001: 302) illustrates, the context of the discourse can have an impact on this diagnostic, which raises some doubt about the reliability of this test. (60)

a.

Der Reinhold kannPERM den Nanga Parbat ohne the Reinhold can the-acc Nanga Parbat without Sondergenehmigung bezwingen. special.permission conquer-inf ‘Reinhold may conquer the Nanga Parbat without special permission.’

b.

Der Nanga Parbat kannPERM vom Reinhold ohne the-nom Nanga Parbat can by.the Reinhold without Sondergenehmigung bezwungen werden. special conquer-ppp pass.aux-inf ’The Nanga Parbat may be conquered by Reinhold without special permission.’

(61)

a.

Der Kräuterpfarrer kannPOSS die Leserbriefe auch the herbs.priest can the-acc letters.to.the.editor also schon mal persönlich beantworten. par sometimes personally answer-inf ‘Sometimes, the herbal priest personally answers the letters to the editor.’

36 DeReKo: M07/DEZ.04402 Mannheimer Morgen, 15.12.2007.

52 | 2 Modal verbs: A class struggle

b.

Die Leserbriefe könnenPOSS vom Kräuterpfarrer the-nom letters.to.the.editor can by.the herbs.priest auch schon mal persönlich beantwortet werden. also par sometimes personally answer-ppp pass.aux-inf ‘Sometimes, the letters to the editor are answered by the herbal priest himself.’

All of these diagnostics illustrate that some of the circumstantial interpretations of können are indeed to be analysed as raising predicates. It seems to be possible that all of these different readings are derived from one general reading. This is supported by diachronic evidence. As Fritz (1997: 14) demonstrates, the permission reading developed out of practical possibility meaning. A similar position is put forth by Öhlschläger (1989: 156), who argues that permission reading is only secondary and is generated by a specific conversational background. As can eb clearly seen, the last remaining reading can also be related to the practical possibility reading. Carlson (1977: 119) and Brennan (1993: 97) discuss instances of the English modal can in which it quantifies over elements that are smaller than worlds. In particular, they focus on quantification over individuals (cf. 536) or situations. These readings can be equally found with its German counterpart können in (63) and (64): (62)

A basketball player can be short.

(63)

Ein CSU-Politiker kann evangelisch sein. a CSU.politician can protestant be-prf.aux.inf ‘A CSU politician can be a protestant.’

(64)

Ein CSU-Politiker kann auch durchaus mal die Ehe a CSU.politician can also sometimes once the marriage gebrochen haben. break-ppp have-inf ‘There are some CSU politicians who have committed adultery.’

(65)

Ein derartiger Blutwert kann genetisch bedingt, aber a such blood.value can genetically determine-ppp but auch auf Doping mit EPO oder Eigenblut zurückzuführen sein also on doping wiht EPO or own.blood to.backtrack-inf be-inf – derartige Dopingfälle gab es im Langlauf bereits such doping.cases gave it in cross.country already zuhauf.³⁷ in.masses

2.2 Case studies |

53

‘Such a blood parameter can be congenital but also due to doping with EPO or an autologous transfusion – in cross country, countless such cases of doping have already occurred.’

(66)

Ein aktiv gemanagter Fonds kann aber durchaus besser an actively managed fund can yet definitely better abgeschnitten haben. Sie federn extreme Bewegungen ab und perform-ppp have-inf they absorb extreme movements off and haben vielleicht nur 15 bis 18 Prozent verloren. Viele have may only 15 to 18 percent lost many Fondsmanager haben seit Jahren gezeigt, dass ihre aktiv fonds.managers have since years shown that their actively gemanagten Fonds besser laufen als passive.³⁸ managed funds better run than passive ‘A fund which is actively managed can have performed better. They absorb extreme price movements and they have possibly lost only 15–18 percent. Many managers of funds have demonstrated throughout the last years that their actively managed fonds perform better than those that are managed passively.’

(67)

Drei ihrer Künstler hat die Galeristin bereits während ihres three her-gen artists-gen has the gallery owner during their Studiums an der Mainzer Kunsthochschule kennen gelernt. „Ein studies at the Mainz Art.school know learnt a Künstler kann auch schon mit 30 einen Formenkanon gefunden artist can also already with 30 a form.canon find-ppp haben, den er entwickeln will.”³⁹ have-inf that he develop-inf wants ‘The gallery owner met three of her artists during their studies at the art school in Mainz: “An artist can have found his style already at the age of 30.” ’

(68)

Wer danach dieselben Türen berührt und dann ein Eis isst, who after the.same doors touches and then a ice.cream eats kann sich schon angesteckt haben.⁴⁰ can self already infect-ppp have-inf ‘Some of those who touched the same doors and went to eat ice cream can already have been infected.’

Essentially, the semantic contribution of können is that it existentially quantifies over the sort of individual denoted by the subject NP: There are at least some CSU

37 DeReKo: NUZ06/OKT.02424 Nürnberger Zeitung, 25/10/2006. 38 DeReKo: M08/SEP.70220 Mannheimer Morgen, 09/09/2008. 39 DeReKo: RHZ07/APR.07409 Rhein-Zeitung, 10/04/2007. 40 DeReKo: RHZ09/OKT.00336 Rhein-Zeitung, 01/10/2009.

54 | 2 Modal verbs: A class struggle

politicians for which the property encoded by the predicate holds. In contrast, the practical possibility readings discussed above seem to be the same phenomenon that Brennan (1993: 97) has in mind when she talks about quantification over situations. Some authors, such as Portner (2009: 134), suggest a different classification, according to which quantificational modals are a subtype of dynamic modality. Summing up, there is actually no need to argue about the existence of circumstantial modal verbs with raising patterns. First of all, there is a lot of empirical evidence ranging from the selection of non-referential subjects to the scope ambiguity of quantifiers, and finally to the transparency with respect to voice. Moreover, it has been demonstrated that even the often quoted main proponents of the control versus raising approach, Jackendoff (1972: 105), Stechow and Sternefeld (1988: 446) and Brennan (1993: 27) acknowledge that there are cases of circumstantial modals with underlying raising pattern. Furthermore, it has been shown that their approaches are never exhaustive and comprise only one or two modals. Therefore, it should not be all too surprising if they reach very different conclusions about the nature of circumstantial modals. Yet, it is not clear whether there are more circumstantial modals with control patterns apart from the obvious cases such as the ability reading of können, and the volitional verbs wollen, möchte and mögen. As Feldman (1986: 179) discusses, deontic modality comprises at least two subtypes, whereas ‘the ought-to-do involves a relation between an agent and a state of affairs. The ought-to-be involves a property of state of affairs’. In a similar fashion, Abraham (2001: 18, 2002: 36 2005: 241, 257, 261) and Barbiers (2002: 67) argue that deontic modals such as the permission reading of können are generally possible as control verbs. Brennan (1993: 45) applies a test similar to the voice transparency that is based on symmetric predicates. The result could be interpreted in favour of the existence of deontic control verbs. In a similar vein, Diewald (1999: 62) presents analogous examples of the voice transparency test. Reis (2001: 302), however, relativises the reliability of this diagnostic. 2.2.1.4 Raising directional phrases with event modification As already observed by Paul (1898: 296 §221), the traditional six modal verbs including lassen ‘let’ and helfen ‘help’ systematically embed directional phrases in the absence of a corresponding infinitive. He concludes that the lack of the infinitive cannot be explained in terms of an ellipsis, as has been proposed by some authors, such as Heyse (1822: 403). In a similar vein, Zifonun (1997: 1256), Erb (2001: 94), Vater (2004: 18), Szumlakowski-Morodo (2006), Hetland and Vater (2008: 102) and Ørsnes (2007) argue that the traditional six modal verbs in Ger-

2.2 Case studies |

55

man alternatively select directional PP or AP complements. This phenomenon is not restricted to German but also occurs in other Germanic Languages, such as Dutch (cf. Barbiers (1995) and Barbiers (2002), Riemsdijk (2002)), Danish (cf. Jespersen (1931: 238)), Norwegian (cf. Hetland and Vater (2008: 102)) and, as Wilder (2008: 249) points out, in a number of Slavic languages such as Czech, Slovak, Slovenian, Serbian, Bulgarian and Russian. Furthermore, Denison (1993: 305) has demonstrated that these patterns were also available in in earlier stages of English; Fritz (1997: 72) provides evidence for the existence of them in Old and Middle High German. Barbiers (1995: 151, 2002: 53) presents a whole range of serious obstacles to be overcome for an ellipsis account for Dutch: (i) essentially, it is hard to account for patterns consisting of a inanimate subject + modal verb + directional phrase, and (ii) it is not obvious why modal verbs with verbless directional phrases can never be interpreted epistemically. Szumlakowski-Morodo (2006: 327) presents an analogous example for German: (69)

Der Brief muss zur Post (#gehen) /( #gebracht werden) The letter must to.the post.office go brought pas.aux ‘The letter must be taken to the post office’

Neither of the alternatives is appropriate. The first case is problematic because gehen ‘go’, as used in example (69), prototypically requires an animate subject. Nevertheless, it would in principle be possible to introduce this verb of movement, but it would also subtly affect the semantic interpretation of the utterance. Barbiers (1995: 155) makes a similar observation for Dutch. Even if the second solution is more adequate from a semantic perspective, it is more implausible from a syntactic point of view. First of all, it presupposes the ellipsis of the passive auxiliary, an assumption which lacks empirical justification, since werden cannot be omitted in contemporary German. Moreover, the agent argument in a passivised utterance can usually be expressed by a von-PP. However, this option is not available in example (69), as pointed out by Barbiers (1995: 152). Analogous to Barbiers (1995: 153, 2002: 54), Vater (2004: 18) and Hetland and Vater (2008: 102) observe that modals with verbless directional phrases are restricted to a non-epistemic interpretation. (70)

Hans muss ins Feld. Hans must into.the fieldpas.aux ‘Hans has to go into the field.’ (deontic) ‘*Hans must go into the field.’ (epistemic)

Finally, Szumlakowski-Morodo (2006: 342) discusses a third challenge for an ellipsis account. Whenever one of the traditional six modal verbs selects a verbless

56 | 2 Modal verbs: A class struggle

directional phrase, it will be realised as a ge-participle, rather than as IPP in perfect tense environments, in contemporary Standard German. In some southern varieties, however, the IPP is nevertheless available in these contexts. (71)

. . . dass er nachhause gemusst hat. that he home must-ppp(ge) prf.aux ‘. . . that he had to go home.’

As was pointed out in Section 2.1.1.2, the ge-participle is ungrammatical if the traditional six modal verbs embed an infinitive. Therefore, it is fairly likely that the use with a verbless directional phrase in example (71) must be something else. Since an analysis of verbless directional phrases in terms of ellipsis turns out to be inadequate, an alternative explanation becomes necessary. Paul (1898: 296 §221) already explicitly suggested that directional phrases can be considered as predicates:⁴¹ ’[...] er ist weg, er ist nach Rom, die nicht anders aufzufassen sind als er ist in Rom, dh. weg und nach Rom sind als Prädikate zu nehmen ist als Kopula. Desgleichen er ist von Rom, woher ist er?.

In a similar vein, Barbiers (1995: 162, 2002: 57) and Erb (2001: 95) assume that the verbless directional phrases under discussion can be analysed as small clause complements; Ørsnes (2007) develops an analogous solution within LFG. There have only been few attempts to develop an analysis that assumes ellipsis of the infinitive. As already shown in Section 2.2.1.1, the modal verbs that lack infinitive complements differ crucially from canonical ellipsis. As pointed out by Ørsnes (2007: Sections 3 & 4.2), the only solution would be to assume an entry of a specific empty verb in the lexicon. This is exactly the solution Riemsdijk (2002: 187) opts for. He argues that, in virtue of being functional categories, modal verbs in Germanic languages can licence the empty light verb GO. Wilder (2008) adapts this proposal and extends it to other languages and related phenomena. From the discussion above, it has become clear that any account that assumes an empty verb faces a whole range of challenges. Some of them can be circumvented by means of a number of theory-specific stipulations, but in the end Riemsdijk (2002: 166) concedes that his approach cannot explain why modal verbs with verbless directional phrases are always interpreted circumstantially. Moreover, he explicitly restricts his analysis to modal verbs, cf. Riemsdijk (2002: 144). In contrast,

41 [...] er ist weg ‘he is away’, er ist nach Rom ‘he is to Rome’, they have to be interpreted in the same way as er ist in Rom, ‘he is in Rome’, accordingly weg ‘away’ and nach Rom ‘to Rome’ have to be considered as predicates, ist ‘is’ as copula. In a similar fashion, er ist von Rom ‘he is from Rome’, woher ist er? ‘where is he (from)?’ [own translation]

2.2 Case studies |

57

Öhlschläger (1989: 64) assumes that the verbless directional phrases are the result of an indefinite ellipsis, as suggested by Shopen (1973: 68). In Shopen’s (1973) proposal, however, indefinite ellipsis includes cases in which the elided element cannot be recovered from the ongoing discourse, such as the sentence ‘John received a package (from Canada)’. But Shopen (1973: 65) explicitly points out that indefinite ellipsis is a subtype of constituent ellipsis, ‘where a predicate is expressed without all its arguments’. This phenomenon only concerns cases in which an entire argument is elided. Yet, this does not hold for modal verbs that occur with verbless directional phrases. Adopting the analysis suggested by Shopen (1973), we would expect the whole infinitival VP to be suppressed, including its head, the motion verb and the dependent directional phrase. This actually underpins Öhlschläger’s original claim. Finally, Behaghel (1924) does not become explicit as to whether he considers an account in terms of ellipsis an appropriate solution. In some passages, he seems to favour an ellipsis account (p. 369) in others he seems to adopt Paul’s (1898) predicate analysis (p. 179).⁴² Since Paul (1898: 296 §221) already demonstrated that the selection of verbless directional phrases is not limited to the traditional modal verbs, any solution should also be applicable to the remaining verbs that come into consideration. However, a couple of them cannot be captured by an analysis in the spirit of Riemsdijk (2002). On the one hand, there are verbs like helfen ‘help’. From a semantic perspective, a predicate like the empty GO suggested by Riemsdijk (2002) seems to be roughly compatible with the directional phrase in (72). But for some strange reason, helfen loses its usual ability to alternatively combine with bare infinitive complements once it selects a directional phrase. In case the speaker wants to express a non-finite complement, the infinitival marker zu becomes obligatory. This is remarkable, since usually, helfen alternatively selects bare infinitives and zuinfinitive complements, as has been shown by Askedal (1989: 5). In order to maintain an ellipsis account, it would become necessary to additionally assume the existence of an empty infinitive marker zu, or the existence of an empty zu-infinitive. On the other hand, there are verbs such as bekommen ‘get’ and kriegen ‘get’ that occasionally select verbless directional phrases or verb particles in an ana-

42 Behaghel (1924: 179): Es versteht sich, daß auch die Hilfszeitwörter behandelt werden, die infolge der Ersparung eines Bewegungsverbs selbst dessen Bedeutung erhalten haben. (‘It is clear that the auxiliaries will also be treated, which have acquired the meaning of a predicate of motion due to the deletion of this predicate’ [own translation].) Behaghel (1924: 369): Wenn die Ergänzung des Hilfszeitwort erspart wird, so kann das Partizip die Infinitivform beibehalten oder die echte Partizipform annehmen. (If the complement of an auxiliary is deleted, the participle can keep its form as an infinitive or adopt the genuine form of a participle [own translation].)

58 | 2 Modal verbs: A class struggle

logous function, as is illustrated by examples (73)–(76). In Standard German, it is not possible to insert any verb after the directional phrase. Some northern varieties allow for that option, but in that case the meaning will always be more specific in comparison to the variant without the verb of motion: (72)

Ich helfe Dir auf den Baum (zu kommen) (zu #gehen) I help you-dat on the table to come-inf to go-inf ‘I help you to climb the tree’

(73)

Neunzig Minuten lang bekam keiner den Ball ins Tor ninety mintes long got nobody the ball into goal (*gegangen)/ (*gekommen).⁴³ go-ppp(ge)/ come-ppp(ge) ‘During the ninety minutes, nobody managed to get the ball into the goal.’

(74)

Er kriegte ihn nicht hoch, aber liebte mich.⁴⁴ he got him neg up but loved me ‘He could not get it up, but he loved me.’

(75)

Alle kriegen ihren besten Freund hoch, nur nicht der all get their best friend up only neg the bedauernswerte Victor-Emmanuel Chandebise.⁴⁵ unfortunate Victor-Emmanuel Chandebise ‘Everybody could get it up but the unfortunate Victor-Emmanuel Chandebise could not.’

(76)

Paul kriegt keinen Ton heraus, er hat längst keine Stimme Paul gets no sound out he has long no voice mehr.⁴⁶ anymore ‘Tom can’t get a sound out; he lost his voice a while ago.’

To conclude, in view of the evidence presented above it is not plausible to assume that modal verbs that occur with verbless directional phrases involve ellipsis of a predicate of motion. The most revealing fact is that these types of complements are restricted to modals with circumstantial and or dynamic interpretation only. This also holds at least for the remaining Germanic languages, as shown by Hetland and Vater (2008: 102) for Norwegian and Barbiers (1995) and Barbiers (2002) for Dutch. This is hard to account for with an analysis that assumes ellipsis.

43 44 45 46

DeReKo: RHZ05/OKT.11553 Rhein-Zeitung, 10/10/2005 DeReKo: 97/SEP.22636 Zürcher Tagesanzeiger, 19/09/1997; Allen Ginsbergs letzte Worte. DeReKo: K00/JUL.55029 Kleine Zeitung, 23/07/2000. DeReKo: RHZ97/DEZ.00540 Rhein-Zeitung, 01/12/1997.

2.2 Case studies |

59

Assuming that directional phrases can be interpreted as predicates, a solution can easily be obtained. According to the rule for coherence formulated by Bech (1955: 65), verbs that sub-categorise bare infinitive complements or past participles obligatorily form a predicate complex (verbal cluster). Interestingly, almost all verbs of this class may alternatively select verbless directional phrases. In contrast, motion verbs occasionally take bare infinitive complements or, in rare cases, past participles, as was shown in Section 2.1.2.1: – verbs selecting bare infinitives: können, müssen, dürfen, sollen, mögen, brauchen, lassen ‘let’, würde gerne ‘would like to’, helfen ‘help’, trauen ‘dare’, haben ‘have’, sein ‘be’, tun ‘do’ – verbs selecting past participles: gehören ‘belong/should.be’, bekommen ‘get’, kriegen ‘get’, sein ‘be’, haben ‘have’ – motion verbs selecting bare infinitives: kommen ‘come’, gehen ‘go’, schicken ‘send’, senden ‘send’ – motion verbs selecting past participles: kommen ‘come’, bringen ‘bring’⁴⁷ As it turns out, there are three types of predicates in German that always trigger predicate complex formation: Those which select for bare infinitives, past participle or directional phrases. It is evident that each of them involves a particular semantic specification. Obviously, only bare infinitives fulfil the prerequisites for an epistemic interpretation, as observed by Reis (2001: 310). Possibly, the crucial property of bare infinitives is to encode states. As Barbiers (2002: 59) illustrates, verbless directional phrases always denote a polarity transition and correspond rather to perfective or eventive predicates. Of course, as already shown by Bech

47 Vogel (2005) discusses the pattern kommen + past participle. However, this phenomenon does not seem to be restricted to kommen. In addition to the patterns discussed by Vogel (2005), verbs of caused movement, such as bringen ‘bring’, are observed in analogous patterns. (1)

Soldaten bringen den Bauern geschleppt. soldiers bring the farmer drag-ppp ‘Soldiers drag the farmer along.’ Friedrich von Schiller Wallensteins Lager, 10. Auftritt (1799).

(2)

Was bringst’n da wieder geschleppt, sag emal? what bring.part there again drag-ppp say once ‘An’ what’ that you’ve got hold of now?’ Gerhart Hauptmann Die Weber V (1893).

At this point, I wish to thank Peter Sprengel, who helped me with some puzzles in the translation from Silesian into German. A similar example is provided by Schoetensack (1856: 293).

60 | 2 Modal verbs: A class struggle

(1955: 84), there are also verbs that select zu-infinitive complements and nevertheless obligatorily form a predicate complex, such as brauchen or pflegen ’be wont to do something’. Moreover, Ørsnes (2007: Fn. 23) points out that there is also a small group of object control verbs that alternate between zu-infinitive complements and verbless directional phrases, such as bitten, wünschen and schaffen. Since an exact description of the relation between the small group of zu-infinitives and verbless directional phrases goes beyond the goals of the present investigation, it will be the task of some future enterprise to precisely examine the interdependence between the two classes of predicates. When adopting a small clause or a predicate analysis, one more problem remains to be tackled. Verbs that embed a bare infinitive complement can be either classified as control verbs, or as raising verbs, depending on whether or not they carry a subject argument of their own. Assuming that directional phrases are predicates, this issue also needs to be addressed here. Since directional predicates do not usually combine with non-referential subjects, one major diagnostic for raising cannot be applied here (though it seemed to be possible in earlier stages of German, as will be shown in Section 2.2.6.3). It might be revealing to verify to what extent inanimate subjects are compatible with directional predicates. In the case of können with a possibility reading, this type of subject appears to be acceptable. It is not plausible to assume that the possibility is ascribed to the sun in example (78), resulting in some kind of ability reading. Rather, the possibility operator takes scope over the whole predication ‘The heat of the sun (gets) in’. In all examples of können with verbless directional phrases provided so far by Vater (2004), Hetland and Vater (2008), Szumlakowski-Morodo (2006), Ørsnes (2007) and Erb (2001: 96), this verb denotes a possibility or a permission. Since the two corresponding counterparts of können with bare infinitive complements involve raising rather than control, it is fairly likely that this is also the case in the examples (77) to (79). As illustrated in example (80), there are some rare instances of können with a verbless directional phrase in which an ability interpretation is most plausible. As shown in Section 2.2.1.2, this in turn requires the presence of a subject argument which is an indicator for control. (77)

Waffen zu Hause müssten stärker kontrolliert werden, aber die arms at home must stronger controlled be but the Polizei kann nicht ohne Weiteres in die Wohnungen hinein.⁴⁸ police can neg with.out further into the appartment in

2.2 Case studies |

61

‘More strict controls should be in place concerning the possession of fire arms in the house the police are not allowed to enter without further permission.’

(78)

Die Sonnenwärme kann hinein, aber nicht wieder heraus.⁴⁹ the sun.heat can in but neg again out ‘The heat from the sun can get in but it cannot get out.’

(79)

Es sollte gewährleistet sein, dass sie jederzeit in Haus oder it should warranted be that they always in house or Wohnung hinein können.⁵⁰ appartment in can ‘It should be ensured that cats can get into the house or apartment anytime.’

(80)

Ich kann schon alleine auf´s Katzenklo und erkunde I can already alone at.the cat.litter.pan and explore gerade die große weite Welt.⁵¹ currently the great wide world. ‘As a kitten, I can already find the cat litter tray on my own and currently I am about to start exploring the big wide world.’

This section demonstrated that verbless directional phrases that occur with the six traditional modal verbs are not a result of ellipsis of a motion verb. Rather, they have to be considered as predicates or small clauses. As is clear from the examples above, these patterns can in principle involve raising as well as control. The reason why Barbiers (1995: 162) argues that directional phrases always induce raising is inherent to the Government & Binding theory and not mandatory for any other type of theory. 2.2.1.5 Raising infinitives with propositional modification Epistemic modifiers are characterized by three important properties. First of all, they do not encode statements about the actual world, but rather express assumptions about possible worlds. Secondly, they do not indicate that the speaker has knowledge of factive realisations of the predication. A phrase like ‘the lake could be cold’ can even be uttered if the speaker does not know whether there was a single moment at which the lake was indeed cold. Finally, a proposition in the

48 DeReKo: BRZ09/MAI.05378 Braunschweiger Zeitung, 12/05/2009. 49 DeReKo: RHZ09/FEB.09586 Rhein-Zeitung, 11/02/2009. 50 DeReKo: BRZ09/JAN.03341 Braunschweiger Zeitung, 09/01/2009; Was Hunde- und Katzenbesitzer zurzeit beachten sollten. 51 http://www.neue-mitmach-zeitung.de/ostrhauderfehn/lokales/katzenfindelkinder-inostrhauderfehn-wer-hat-ein-herz-d931.html, 28/01/2010.

62 | 2 Modal verbs: A class struggle

scope of an epistemic modal operator can never be part of the speaker’s knowledge. Similar observations have already been made by Westmoreland (1998: 12), Diewald (1999: 209, 225) Ziegeler (2006: 90), Fintel and Gillies (2010: 353), Kratzer (2011, 2012: 99) and Martin (2011). Note that there is a similar claim by Zimmermann (2004: 256) about the German discourse particle wohl. This section deals with the epistemic interpretation of können. The main characteristic of epistemic modifiers is that they are evaluated with respect to the knowledge of the speaker (deictic centre). Adopting the position developed by Westmoreland (1998: 12) and Ziegeler (2006: 90), the use of an epistemic modifier presupposes that the speaker (deictic centre) does not know whether the epistemically modified proposition holds or not. For the sake of simplicity, this condition was labelled Condition on Deictic Centres (CoDeC) in Section 2.1.3.3. Accordingly, this section will only deal with instances of what Lyons (1977: 797) defines as ‘subjective’ epistemicity. As will be shown in Section 4.22, all of the examples for objective epistemicity discussed so far can be considered either as subtypes of circumstantial modality such as practical necessity or practical possibility, or as (‘subjective’) epistemic modality. Some authors, such as Palmer (1986: 53), have a broader definition of epistemic modality, which encompasses judgements (speculative, deductive) and evidentials (reportative, based on senses). The term epistemicity as it is used here only comprises Palmer’s (1986) first subtype: Judgements. According to the CoDeC, a speaker (deictic centre) who uses an epistemic modal verb signals that the epistemically modified proposition is not part of his knowledge. So whenever a speaker knows that the proposition under discussion is true, the modal verb employed cannot be an epistemic one. In this section, the extent to which propositions in the scope of an epistemic modal verb may be part of the speaker’s knowledge will be tested. Epistemic modal verbs can be characterised in terms of two types of environments: (i) environments in which they are possible while their non-epistemic counterparts are excluded, and (ii) environments in which they are excluded while their non-epistemic counterparts are possible. As will be illustrated in more detail in Chapter 3, circumstantial modal verbs with referential subjects fail to embed predications that include an identified individual and a predicate that refers to an event in the past or a predicate that denotes permanent states that cannot be changed. Epistemic modal verbs crucially differ in this respect. They are even highly frequent in such environments, as is illustrated in example (81) and (82). Similar observations have been made by Barbiers (1995: 148) and Barbiers (2002: 59). Likewise, Bech (1949: 43) already wonders why modal verbs with infinitive perfect complements target the “reality” of the embedded predication rather than its “realisation”.

2.2 Case studies |

(81)

63

RUPRECHT: [...] Es kann ein dritter wohl gewesen sein.⁵² it can a third perhaps be-ppp inf RUPRECHT: ‘Perhaps, it could have been a third person.’

(82)

Nach Zeugenangaben könnte ein etwa 30 Jahre alter Mann after witness.reports can-sbjv.pst a about 30 year old man der Täter sein.⁵³ the culprit be-inf ‘According to witness reports, the culprit could be a man who is about 30 years old.’

Crucially, in both examples above the proposition in the scope of the epistemic modal verb cannot be part of the speaker’s knowledge. Neither does Ruprecht know who has broken the jug at the moment of his utterance, nor does the author of the other sentence know that the about 30 year old man is the culprit. This is an essential contrast to practical possibility modal verbs and quantificational modal verbs that are in principle compatible with utterance situations in which the speaker can know that the embedded predication holds, as will be demonstrated in more detail in Chapter 3. The second way to characterise epistemic modal verbs is in terms of the environments from which they are excluded, while their circumstantial counterparts are fully acceptable. In the past decades, more than twenty such contexts have been suggested. Based on large data samples exploited from the DeReKo corpus, it will be demonstrated in some detail in Chapter 4 that only a couple of them are empirically justified: Epistemic modal verbs in German do not occur with verbless directional phrase complements, they cannot be separated from their infinitive complements in wh-clefts, they do not undergo nominalisation, they are banned from adverbial infinitives, and finally, they cannot occur embedded under another modal operator. These results are similar to the conclusion that Eide (2005: 9) arrives at for Norwegian. It deserves closer attention that epistemic können comes in two different guises: As kann, with indicative inflection, and as könnte, with subjunctive past inflection. As it turns out, the two alternatives are not interchangeable because they differ with respect to some semantic subtleties. The main question that arises at this point is what the semantic effect of subjunctive morphology on epistemic modal verbs is. As has been observed by Lötscher (1991: 347), epistemic modal verbs that occur in the scope of a subjunctive past marker behave in an unexpected way. Based on the assumption that the past subjunctive in German indicates the counterfactuality of the modified proposition in non-embedded environments,

52 Heinrich von Kleist Der Zebrochene Krug, 9. Entry (1806). 53 DeReKo: RHZ08/JUN.01622 Rhein-Zeitung, 03/06/2008.

64 | 2 Modal verbs: A class struggle

one would expect that an epistemic possibility in the scope of a subjunctive past operator would be interpreted as counterfactual epistemic possibility. However, this is clearly not the case in examples like (82): The speaker does not want to express that the epistemic possibility for the culprit to be the 30 year-old man is banned from the actual world. In more detail, Lötscher (1991: 347) argues that kann and könnte can be mutually replaced without causing any semantic effect. A similar observation has been made by Coates (1983: 239), who argues that the subjunctive mood in English, by and large, leaves epistemic modals unaffected. In contrast, Fritz (1997: 101) observes that subjunctive morphology has an impact on the degree of certainty that is expressed by the epistemic modal verbs können and müssen. As he argues, the degree of certainty of these epistemic modal verbs decreases when a subjunctive of the past morpheme is attached. After reviewing corpus examples for epistemic instances of könnte, it turns out that Lötscher’s claim was wrong: The verb forms kann and könnte cannot always be replaced with each other without affecting the overall meaning of the sentence: First of all, a substitution of könnte with its indicative counterpart can yields a decreased degree of acceptability in most contexts, as the contrasts illustrated in (83)–(84) indicate: (83)

a.

Sollte diese Einschätzung zutreffen, dann könnte Iran should this assessment hold than can-sbjv.pst Iran bereits vor acht Wochen mit der Herstellung einer already before eight weeks with the production a-gen Atombombe begonnen haben.⁵⁴ nuclear.bomb begin-ppp have-inf ‘If this assessment turns out to be correct, then Iran could have already started to produce a nuclear bomb eight weeks ago.’

b.

# Sollte diese Einschätzung zutreffen, dann kann Iran bereits

should this assessment hold than can- Iran already vor acht Wochen mit der Herstellung einer Atombombe before eight weeks with the production a-gen nuclear.bomb begonnen haben. begin-ppp have-inf Intended: ‘If this assessment turns out to be correct, then Iran could already have started to produce a nuclear bomb eight weeks ago.’

54 DeReKo: A09/FEB.06422 St. Galler Tagblatt, 23/02/2009.

2.2 Case studies |

(84)

a.

65

Das Mädchen hatte im Garten gespielt und plötzlich Blut the girl had in.the garden played an suddenly blood gespuckt. Die Angst der Mutter: Das Kleinkind spat the fear the-gen mother the toddler könnte Glas verschluckt haben.⁵⁵ can-sbjv.pst glass swallow-ppp have-inf ‘The girl was playing in the garden und suddenly she started spitting blood. The mother was afraid the toddler could have swallowed glass.’

b.

# Das Mädchen hatte im

Garten gespielt und plötzlich Blut the girl had in.the garden played an suddenly blood gespuckt. Die Angst der Mutter: Das Kleinkind kann Glas spat the fear the-gen mother the toddler can glass verschluckt haben. swallow-ppp have-inf Intended: ‘The girl was playing in the garden und suddenly she started spitting blood. The mother was afraid the toddler could have swallowed glass.’

Though the examples with kann in (84b) and (83b) are not completely ungrammatical, they represent a fairly unnatural choice. Obviously, kann requires a very specific context, which is not given here. It generally seems that epistemic könnte is much more flexible with respect to its distribution. In contrast, the use of epistemic kann turns out to be fairly restricted. Nevertheless, there are environments in which epistemic kann cannot be substituted with its subjunctive past counterpart könnte. Most importantly, this concerns environments in which epistemic kann occurs in the scope of negation, as is illustrated in examples (85) and (86): (85)

a.

An mangelndem Training kann es am Montag nicht gelegen on lacking training can it on Monday neg lie-ppp haben.⁵⁶ have-inf ‘The lack of training cannot have been the cause on Monday.’

b.

# An mangelndem Training könnte es am Montag nicht gelegen

on lacking haben. have-inf

training can

it on Monday neg lie-ppp

Intended ‘The lack of training cannot have been the cause on Monday.’

55 DeReKo: BRZ09/AUG.03524 Braunschweiger Zeitung, 08/08/2009. 56 DeReKo: HMP09/AUG.01455 Hamburger Morgenpost, 15/08/2009.

66 | 2 Modal verbs: A class struggle

(86)

a.

Die Tat kann ich nicht begangen haben, zum the crime can I neg commit-ppp have-inf at.the beschriebenen Zeitpunkt saß ich schon einen Tag ein.⁵⁷ given moment sat I already a day in ‘I cannot have committed the crime as I was already in jail at that given moment.’

b.

# Die Tat

könnte ich nicht begangen haben, zum the crime can-sbjv.pst I neg commit-ppp have-inf at.the beschriebenen Zeitpunkt saß ich schon einen Tag ein. given moment sat I already a day in

Intended: ‘I cannot have committed the crime as I was already in jail at that given moment.’

The only interpretation that is available in examples (85b) and (86b) is the one in which the negation is in the scope of the modal verb, yielding readings such as It could be the case that the lack of the training was not the cause and It could be that I have not committed the crime. But these readings are not plausible, in the given context. As the data in the examples given above indicates, there is a subtle but solid difference between the epistemic interpretations of the indicative form kann and the subjunctive past form könnte. But how can this distinction be captured? At least two different types of epistemic könnte have to be considered. The example that involves a conditional (cf. 83a) seems to be revealing. In this particular instance, the speaker indicates that his evidence involves a premise that is not verified. In this conditional construction the non-verified premise is explicitly realised as the antecedent: Sollte diese Einschätzung zutreffen (‘If this assessment is correct’). As the conditional in this example is an epistemic conditional (cf. Section 3.4.1), the consequent of this conditional expresses a proposition that is not part of the speaker’s knowledge, an epistemically modified proposition. The role of the epistemic modal verb in this context appears to be redundant at first glance. But upon closer inspection, it turns out that it obviously specifies the degree of certainty of this assumption. Given these observations, there are some reasons to assume that there is one type of epistemic könnte that refers to a conclusion that is based on evidence involving non-verified premises. According to Kasper (1987: 24–28), the subjunctive of the past indicates that the modified proposition cannot be felicitously asserted. In the canonical case, the proposition is interpreted as counterfactual. As has been shown by Lötscher (1991: 339), a subjunctive past operator that takes scope over a modal operator does not express the counterfactuality of the modal oper-

57 DeReKo: RHZ09/NOV.18511 Rhein-Zeitung, 20/11/2009.

2.2 Case studies |

67

ator, as would be expected; rather it results in a factual interpretation. Maintaining Kasper’s view, one could assume that a speaker who uses an epistemic modal verb in the scope of a subjunctive operator intends to communicate that he is not in a position to felicitously utter the epistemic modal verb in indicative mood. So a speaker who utters könnte(p) signals that for some reason he is not in a position to felicitously employ kann(p). Accordingly, in most examples it is not the epistemic possibility that is counterfactual, as was already noted by Lötscher (1991: 347). It is plausible to assume, then, that the cause for the speaker’s reservation has to do with the quality of the evidence on which the epistemic conclusion is based. In other words, the subjunctive morpheme on the epistemic modal verb indicates that the evidence contains premises that are not verified. Furthermore, this account could provide an explanation for why epistemic könnte is hardly acceptable in the scope of a negation whereas epistemic kann is. Employing the indicative epistemic modal verb kann, the speaker indicates that his assumption is based on premises that are verified facts. When uttering ¬♢(p), he points out that this set of facts contains a premise q that contradicts p. Accordingly, the main reason to use such a pattern is to categorically refute the proposition p. The validity of a proposition can only be challenged, if the objections are based on established facts, rather than on non-verified premises. Because it is based on non-verified premises, könnte appears to be less suitable for such a purpose. As example (83a) involves three different types of modifiers (subjunctive past operator, epistemic conditional operator, epistemic modal operator), the precise interaction of these elements still remains to be investigated in further detail. However, the analysis presented here would account for the observation made by Fritz (1997: 101) and Mortelmans (2000: 205), who notice that subjunctive morphology decreases the degree of certainty expressed by an epistemic modal verb. Moreover, it is supported by other recurrent instances of epistemic könnte, as in examples (87)–(89). (87)

Dieses Mal geht es um fehlende Geldbeträge aus der this time goes it about missing money.amounts out the Klassenkassa. Schnell ergibt sich der Verdacht, dass Heinz class.cash.box quickly result refl the suspicion that Heinz Schimmel, ein Mitschüler, der Täter sein könnte.⁵⁸ Schimmel a classmate the culprit be-inf can-sbjv-pst ‘This time, it is about amounts of money that are missing in the cash box of the class. Quickly, the suspicion arose that Heinz Schimmel, a classmate, could be the culprit.’

58 DeReKo: NON09/NOV.13407 Niederösterreichische Nachrichten, 23/11/2009.

68 | 2 Modal verbs: A class struggle

(88)

Das Blut, das er dabei vergoß, könnte für die Gendarmerie the blood that he thereby spilled could for the police eine heiße Spur sein, denn nun besitzt sie einen genetischen a hot trace be-inf as now possesses she a genetic „Fingerabdruck” des Einbrechers.⁵⁹ fingerprint the-gen burglar-gen ‘The blood that he spilled in doing so could be a fruitful hint for the police, as they now have a genetic ‘fingerprint’ of the burglar.’

(89)

Laut Polizei könnte der Junge der bisher unbekannte according police can-sbjv.pst the boy the hitherto unknown Täter sein, der am Montag oder Dienstag in ein culprit be-inf that on Monday or Tuesday into a Wohnhaus im Küferweg eingebrochen war.⁶⁰ residential.building at.the Küferweg in.break-ppp was ‘According to the police, the boy could be the hitherto unknown culprit that on Monday or Tuesday broke into the residential building at Küferweg.’

Once again, in all of these contexts epistemic kann is very unnatural. An interesting case is the example (89). As has been observed by Stephenson (2007: 490), some adverbials can be used to identify the deictic centre. Following this observation, the utterance in example (89) is ambiguous. In the first interpretation, the adverbial laut Polizei (‘according to the police’) serves as deictic centre. Correspondingly, the assumption expressed by the epistemic modal verb könnte is attributed to the referent contributed by the adverbial, which is the police. In the second interpretation, the epistemic conclusion is attributed to the speaker, who indicates that he is not in a position to felicitously use the epistemic modal verb kann in this environment. The most plausible reason is that his conclusion would be based on premises that are non-verified. In the example above, these premises could be referred to by the adverbial laut der Polizei, which would be interpreted as if the police are right. In some cases, the speaker may even draw his conclusion based on premises that he considers as counterfactual or false, as is illustrated in example (90). A similar example of epistemic könnte is discussed by Mortelmans (2000: 208). (90)

Da schreibt Frau Scherfenberg gleich zu Anfang: Darwin there writes Mrs Scherfenberg already at beginning Darwin habe „den Glauben an den allmächtigen Schöpfergott have-sbjv.prs the faith in the almighty creator.deity

59 DeReKo: V99/JAN.03151 Vorarlberger Nachrichten, 21/01/1999. 60 DeReKo: RHZ07/JUL.04745 Rhein-Zeitung, 05/07/2007.

2.2 Case studies |

69

der Bibel zerschmettert”. Wenn das so wäre bzw. the-gen Bible shattered if this so be-sbjv.pst or gewesen wäre, dann könnte es doch wohl nicht be-ppp be-sbjv.pst then can-sbjv.pst it par maybe neg sein, dass die christlichen Kirchen eben diesen Glauben an be-inf that the Christian church precisely this faith in Gott den Schöpfer nach wie vor – auch 150 Jahre nach God the Creator after as before also 150 years after Darwins Werk – jeden Sonntag öffentlich bekennen – und zwar Darwin-gen work every Sunday openly avow and par die Katholiken, die Protestanten und die Orthodoxen in gleicher the Catholics, the Protestants and the Orthodox in same Weise, von den Muslimen u.a. ganz zu schweigen.⁶¹ manner about the Muslims etc completely to be.silent-inf ‘Mrs. Scherfenberg already claims at the outset that Darwin has shattered the faith in the almighty Lord of Creation. If this were indeed the case, then it could not be the case that Christian churches still openly avow this belief on every Sunday encompassing the Catholics, the Protestants and the Orthodoxes, not to mention the Muslims.’

This example is remarkable for another reason: The epistemic instance of könnte occurs in the scope of negation. As has been shown above, this behaviour is rather atypical of epistemic könnte, though very typical of epistemic kann. This leads to the conclusion that the example above involves an instance of könnte that is compositionally built from epistemic kann and the subjunctive past morpheme. In Section 4.4, further examples of this type will be provided. The compositional interpretation is further confirmed by the fact that it can be replaced with its indicative cognate, together with the copula wäre in the conditional clause, which is also inflected for past subjunctive. This indicates that the use of könnte in example (90) is different from the one in the examples (87)–(89) discussed above, where it refers to an epistemic conclusion which is based on non-verified premises. At this point, it is necessary to distinguish between two particular uses of epistemic könnte: one which is based on evidence that contains non-verified premises, and a second one which is based on premises that are false. Whereas the latter type can be accounted for in terms of a compositional interaction between the epistemic operator and the subjunctive past operator, the precise status of the first type is less clear. One option is to assume that the subjunctive operator affects the validity of the premises on which the epistemic conclusion is based. Yet, it remains to be shown how these elements interact in more detail. Otherwise,

61 DeReKo: NUN09/FEB.01667 Nürnberger Nachrichten, 14/02/2009.

70 | 2 Modal verbs: A class struggle

epistemic könnte would have to be analysed as a non-compositional epistemic modal operator. As pointed out by Fritz (1997: 102), this scenario is rather likely. There are at least two methods that could reveal the true nature of this nonverified premise interpretation of könnte, and that could illustrate to what extent it can be compositionally captured. Firstly, there are two ways of how the subjunctive of the past can be morphologically realised (Zifonun (1997: 1736)): either in a synthetic or in a periphrastic manner. In the case of können, the synthetic variant is könnte, and the analytic variant involves the subjunctive past auxiliary würden, and the infinitive können. If the non-verified premise reading is a result of semantic composition, it is expected to also be available in the analytic alternative. As it turns out, such instances of the periphrastic variant that come into consideration for an epistemic interpretation occur fairly rarely in the DeReKO corpus. (91)

Im Programm der Stadt ist schon längst die Sanierung in.the program the-gen city is already long the renovation des Marktplatzes vorgesehen. Hier könnte man the-gen marked.place-gen planned here can-sbjv.pst one statt dem jetzigen Geröll, als flaches Denkmal eine instead the current boulders as flat memorial a Pflasterung mit farbigem Steinmosaik in Form der Karte paving with coloured stone.mosaic in shape the-gen map Europas gestalten. Politisch würde wohl niemand Europe-gen arrange politically sbjv.pst.aux maybe nobody dagegen sein können.⁶² against be-inf can-inf ‘The renovation of the market place has been already long ago considered in the program of the city. Here one could see a flat paving made out of a coloured stone mosaic in the shape of Europe. In political respect, nobody could disagree.’

The fact that the only occurrence of this in the DeReKo corpus is in the scope of negation merits closer attention. Given that there is a negation, it could not be replaced with its synthetic counterpart könnte. This indicates that the example (91) meets all of the criteria that are typical of epistemic kann. Thus, the only interpretation that is applicable to this pattern is the counterfactual premise interpretation, which refers to a counterfactual epistemic possibility. The non-verified premise reading does not apply. In a similar manner, Mortelmans, Boye and Auwera (2009: 34) have illustrated that the analytic pattern cannot construe the non-verified premise for können in German. In contrast, they demonstrate that in

62 DeReKo: I97/SEP.37816 Tiroler Tageszeitung, 27/09/1997.

2.2 Case studies |

71

Dutch this type of interpretation is available. The analytic pattern zou + kunnen can yield a non-verified premise interpretation. The second way to determine the precise status of the non-verified premise interpretation with respect to compositionality is another replacement test that involves semantically related expressions, such as epistemic adjectives möglich ‘possible’ and notwendig ‘necessary’. If the non-verified premise reading is a result of semantic composition, it is expected to also to be available with epistemic adjectives that are selected by a copula with subjunctive past morphology, in patterns such as es wäre möglich, dass ‘it is-sbjv.pst possible that’ and es wäre notwendig, dass ‘it is-sbjv.pst necessary that’. But if the epistemic modal verb könnte in example (87) is substituted with an epistemic adjective, the meaning of the whole utterance will be affected, such as in example (92). (92)

Schnell ergibt sich der Verdacht, dass es möglich wäre, quickly result refl the suspicion that it possible be-sbjv-pst dass Heinz Schimmel, ein Mitschüler, der Täter ist. that Heinz Schimmel a classmate the culprit is ‘Quickly, the suspicion arose that Heinz Schimmel, a classmate, could be the culprit (under some circumstances).’

Once again, the interpretation in which the epistemic possibility is counterfactual is the preferred one. It is hard to decide whether the non-verified premise interpretation is possible at all in this type environment. The past subjunctive of the copula appears to indicate that the possibility is not actual, under the given circumstances. This results from the two replacement tests strongly suggests that the non-verified premise interpretation is not compositional, or at least it involves a very different mechanism. Unlike epistemic modal verbs that are inflected for indicative, their subjunctive of the past counterparts can embed propositions that are known to be false. (93)

Nach all dem was ich weiß, könnte das der Schlüssel according all that what I know can-sbjv.pst this the key zu Zhannas Büro sein. Er ist es aber nicht. to Zhanna-gen office be-inf he is it but neg ‘According to what I know, this could be the key to Zhanna’s office. But it isn’t.’

Examples like (93) describe a conflict between the evidence drawn from the knowledge, and the external evidence provided by the utterance situation. A similar observation has been made by Copley (2006: 5) for the English modal auxiliary should. If könnte is replaced with its indicative cognate kann, the acceptability significantly decreases. However, it still must be examined if this phenomenon applies to könnte to the same extent in its non-verified premise interpretation, and its counterfactual premise interpretation.

72 | 2 Modal verbs: A class struggle

As already indicated at the outset of this section, the most efficient definitions of epistemic modality impose restrictions on the speaker’s (deictic centre’s) knowledge. The accounts reviewed so far differ with respect to the precise formulation of these restrictions. Some accounts assume that the usage of an epistemic modal operator presupposes that the speaker does not know whether the proposition is true or false. As a consequence, neither p nor ¬p should be part of the speaker’s knowledge. In contrast, Martin (2011:Sect. 3.1) argues that a speaker who uses an epistemic modal operator cannot know that the proposition is false, thus the proposition ¬p must not be part of the speaker’s knowledge. Both analyses cannot neatly account for the example given above (93) without any further stipulation. The analysis developed here is based on the assumption that the proposition in the scope of the epistemic modal operator must not be part of the speaker’s knowledge (CoDeC). From this point if view, the example given above can be accounted for, as the knowledge of the speaker only comprises the proposition ¬(this is the key to Zhannas office) but not the proposition this is the key to Zhannas office. As it seems, one major difference between epistemic kann and könnte concerns the way they qualify their underlying evidence. Whereas the former type indicates that the underlying evidence is entirely made up out of facts, the latter signals that some of the premises are not verified. Yet, there are some open questions. First of all, it is not clear why epistemic kann occurs so rarely. Unlike the other epistemic modal verbs in German, it occurs more often in the scope of negation and in questions, which are rather atypical environments. This has led some authors, such as Öhlschläger (1989: 208), to the conclusion that these instances of kann do not involve genuine epistemic modality, but rather ‘objective’ epistemic modality, which is considered as a less grammaticalised type. At this point, it is not evident whether können involves a true independent epistemic reading at all. As for its English counterpart can, it has been shown on various occasions that it lacks this type of reading, as illustrated by Hofmann (1976: 94), Coates (1983: 85), Sweetser (1990: 62), Brennan (1993: 14) and Drubig (2001: 43), for instance. Further details are discussed in Section 3.3. It is a serious matter to decide how to distinguish between an epistemic possibility interpretation, a practical possibility interpretation and a quantificational reading, as their communicative effect can be the same. However, the possibility modal verb kann differs from its English counterpart can in two essential respects: It embeds predications that involve an identified individual, and a predicate that refers to a permanent state or that refers to the past, and in these environments the embedded proposition is not part of the speaker’s knowledge, as the examples below indicate. Both properties are characteristic for genuine epistemic modal operators. Furthermore, it will be demonstrated in Chapter 4 that the assumption of

2.2 Case studies |

73

an ‘objective’ epistemic modality is misleading, thus refuting Öhlschläger’s objections. There are a couple of environments in which it becomes far more likely that können is epistemically interpreted. As Doitchinov (2001: 119) argues, the pattern es kann sein, daß ‘it could be that’ forces an epistemic interpretation, as in examples (94)–(97). Furthermore, modal verbs are preferably epistemically interpreted whenever they select individual-level predicates, as illustrated in examples (98)–(99), or complements with past reference, as shown in examples (101)–(102). A detailed discussion of these diagnostics will be given in Chapter 3. (94)

Es kann sein, dass die Zisterne so um 330 nach Christus hier it can be that the cistern par about 330 after Christ here angelegt worden ist.⁶³ built pas.aux-ppp is ‘It is possible that the cistern was built here about 330 years after Christ.’

(95)

Es kann sein, dass zwischen dem Teamchef und Ivanschitz it can be that between the team.leader and Ivanschitz etwas vorgefallen ist.⁶⁴ something happened is ‘It is possible that something happened between the team leader and Ivanschitz.’

(96)

Es kann sein, dass mich die neue Frisur ein wenig schneller und it can be that me the new haircut a little faster and besser gemacht hat [. . . ]⁶⁵ better made has ‘It is possible that my new haircut enabled me to be faster and better . . . ’

(97)

Es kann sein, dass Ardi ein direkter Vorfahr ist.⁶⁶ it can be that Ardi a direct ancestor is ‘It is possible that Ardi is a direct ancestor.’

(98)

Vieles spricht dafür, dass der festgenommene Mann der Täter much speaks in.favour that the arrested man the culprit sein kann.⁶⁷ be can ‘There are many factors in favour of the assumption that the arrested man could be the culprit.’

63 64 65 66 67

DeReKo: RHZ08/JUL.05907 Rhein-Zeitung, 04/07/2008. DeReKo: NON09/NOV.05778 Niederösterreichische Nachrichten, 10/11/2009. DeReKo: HAZ09/DEZ.03174 Hannoversche Allgemeine, 21/12/2009. DeReKo: NUZ09/OKT.00590 Nürnberger Zeitung, 08/10/2009. DeReKo: RHZ08/FEB.11333 Rhein-Zeitung, 13/02/2008.

74 | 2 Modal verbs: A class struggle

(99)

Diese traditionelle Schilderung kann allerdings auch falsch sein.⁶⁸ this traditional description can however also false be ‘However, this traditional description could also be false.’

(100)

So kann die Motte in Wipshausen einmal ausgesehen haben.⁶⁹ so can the Motte in Wipshausen once out.look-inf have-inf ‘The Motte in Wipshausen may have looked like this once upon a time.’

(101)

Ich kenne den Täter nicht, er kann die Taten auch begangen I know the culprit neg he can the acts also committed haben, um Macht und Kontrolle auszuüben.⁷⁰ have in.order.to power and control exert-inf ‘I don’t know the culprit, but he may have also committed the acts in order to exert power and control.’

(102)

Es kann auch ein zusätzlicher Einsatz-Alarm das Signal it can also a additional mission.alarm the signal überlagert haben.⁷¹ interfere-ppp have-inf ‘There could have been an an additional alarm that interfered with the signal.

Note that in none of the examples could the speaker resume the discourse by stating ‘. . . and I know that is the case’. Doitchinov (2001) is not quite right in his claim that es kann sein, daß needs to be obligatorily interpreted in an epistemic way, since there are rare cases in which the speaker could in principle know whether the proposition holds or not: (103)

Die Zahnradstrecke mit bis zu 90 Promille Gefälle verbindet the cog.railroad with up to 90 per.mill slope connects mitunter zwei völlig unterschiedliche Klimazonen: Es occasionally two completely different climes: it kann sein, dass in Heiden tiefster Winter ist, am See unten can be that in Heiden deepest winter is at.the lake down jedoch alles grün. Oder umgekehrt: In Rorschach herrscht however everything green Or vice.versa in Rorschach reigns neblige Kälte, in Heiden dagegen warmes, sonniges Wetter⁷² foggy cold in Heiden in.contrast warm sunny weather

68 DeReKo: WPD/AAA.01884 Leipnizkeks, Wikipedia, 2005. 69 DeReKo: BRZ09/AUG.04565 Braunschweiger Zeitung, 11/08/2009. 70 DeReKo: HAZ09/FEB.00785 Hannoversche Allgemeine, 05/02/2009. 71 DeReKo: NON09/DEZ.05190 Niederösterreichische Nachrichten, 08/12/2009.

2.2 Case studies |

75

‘The train track with 90 per mill gradient occasionally connects two completely different weather climates. It can be that it is winter in Heiden and at the lake it is green. And vice versa: It is cold in Rorschach and sunny in Heiden.’

The pattern in example (103) turns out to be an instance of practical possibility or event quantification. The copula sein is interpreted as vorkommen ‘occur’, resulting in a quantification over events: ‘Sometimes it is the case that it is cold, sometimes not’. As a consequence, the speaker could add something like: ‘And by the way, now it is indeed very cold in Heiden’. Summing up, in this section it has been pointed out that epistemic können comes in two major guises. First, there is epistemic kann bearing indicative morphology. It is fairly rare, and it indicates that the underlying evidence on which the epistemic conclusion is based entirely consists of premises that are facts. In contrast, epistemic könnte with a subjunctive of the past inflection is much more frequent. It is found with two interpretations: In its counterfactual premise-reading, it indicates that some of the underlying premises are counterfactual. Moreover, the epistemic possibility is counterfactual. Thus, in terms of compositionality, it behaves exactly as is expected of an epistemic operator in the scope of a subjunctive past operator. Apart from this, könnte can be frequently found with a non-verified premise-reading. In this interpretation, the epistemic conclusion is based on evidence that involves premises that are not verified. This indicates that epistemic modal verbs differ with respect to how they qualify the underlying evidence, and it confirms the observation made by Copley (2006: 11), who illustrated that epistemic should is restricted to temporally remote evidence.

2.2.2 müssen The semantic range of müssen encompasses the expression of a physical need, an obligation, a practical necessity and an epistemic necessity. Depending on its specific meaning, it is subcategorised for a control infinitive, for a raising infinitive or for a verbless directional phrase. Similarly to können, müssen also occurs as quantificational modal verb. Since this only affects those cases in which it is in the scope of negation, it has attracted no attention in research so far. In these uses, it will be interpreted as a universal quantifier in the scope of negation (¬∀) that quantifies over the type of individual encoded by the subject argument.

72 DeReKo: A08/JUL.06635 St. Galler Tagblatt, 28/07/2008.

76 | 2 Modal verbs: A class struggle

2.2.2.1 Control infinitives with event modification In German, müssen is used to express sensations that originate in physical needs related body functions whose initiation are typically beyond control, such as laughing, sneezing, urinating, defecating and vomiting. These uses have already been identified as independent interpretations by Becker (1836: 181) as physical necessity readings. Since the necessity originates within the subject referent in all of these cases, Erb (2001: 78) argues that in these instances, müssen involves a dynamic modality. Similarly to können, it has to be considered a control verb. (104)

Clara mussDYN niesen.⁷³ Clara must sneeze-inf ‘Clara needs to sneeze’

(105)

Zwei Zuschauer müssen lachen. two spectators must laugh-inf ‘Two spectators have no other choice but to laugh / two spectators are bursting out laughing.’

This variant of müssen typically selects intransitive verbs which express a body function of their subject referents, and which therefore select animate subject arguments. As a consequence, the standard diagnostic for control, such as the voice transparency test, cannot be applied. Being restricted to a small class of predicates with experiencer subject arguments, there is no meaningful environment in which it could embed an infinitive that lacks a referential subject such as regnen ‘rain’ or impersonal passives. But this selectional restriction could also be an indication that in its physical need reading, müssen is not compatible with non-referential subjects. The only diagnostic for control that applies is the unavailability of de dicto readings with quantified NPs in subject positions: The quantifier two in example (105) always takes scope over the necessity operator. The canonical interpretation is one in which the two spectators are each affected by a different neurobiological stimulus. An interpretation in which the quantifier is interpreted within the scope of the necessity operator would be possible in a scenario in which one neuro-biological stimulus may affect several bodies at the same time. Since in reality neuro-biological stimuli do not transgress the boundaries of a body, a de dicto reading is not plausible for any quantified subject NP that agrees with an instance of müssen that encodes a physical need. To the extent that these observations hold, there is no reason to consider the physical need reading as a raising construction. Thus, this variant of müssen has to involve a control infinitive and selects a subject argument of its own that en-

73 As cited in Erb (2001: 78).

2.2 Case studies |

77

codes the source of modality. This is supported by evidence from Dutch. As Barbiers (1995: 155) argues, the Dutch counterpart moeten even occurs as a transitive verb in these contexts: (106)

Jan moet en plas. Jan must a wee ‘Jan must go for a wee.’

As was already indicated in the preceding section, it is far from clear whether there are other interpretations of müssen that could involve a control pattern. There is a debate concerning the extent to which deontic modal verbs are control verbs or not. Prominent positions are defended by Ross (1969: 86), Jackendoff (1972: 102), Öhlschläger (1989: 105), Palmer (1990: 47), Brennan (1993: 25), Barbiers (1995), Wurmbrand (2001) and Abraham (2001: 18, 2002: 36, 2005: 241, 257, 261). By means of the diagnostics invoked here, it is easier to prove that a verb is a raising verb than proving that it is a control verb. As will be shown, it is evident that there are deontic instances of müssen that are raising verbs, but it is not a trivial task to find compelling evidence for the existence of deontic modal verbs that are control verbs. This issue will be addressed in more detail in Section 2.2.2.3. 2.2.2.2 Control directionals with event modification Likewise, the physical need reading can also be found with verbless directional phrases. (107)

Ich muss aufs Klo. I must at.the toilet ‘I need to go to the toilet.’

Again, it is evident that the modal force originates from within the subject referent. Thus, it fulfils the criterion of dynamic modality and, as a consequence, the subject NP of müssen has to be its own argument in these cases rather than an argument that has been raised from an embedded predicate. 2.2.2.3 Raising infinitives with event modification As has been demonstrated by Welke (1965: 71), Höhle (1978: 81), Öhlschläger (1989: 105), Palmer (1990: 47), Geilfuß (1992), Kiss (1995: 163), Axel (2001: 40), Reis (2001), Erb (2001: 73) and Wurmbrand (1999, 2001: 201), the German necessity modal verb müssen behaves like a logical one-place operator or a raising verb in its circumstantial interpretation. There are at least three different meanings for circumstantial müssen with a raising pattern that need to be distinguished: The de-

78 | 2 Modal verbs: A class struggle

ontic obligation reading, the practical necessity reading and the quantificational reading, which will be discussed at the end of this section. These interpretations differ with respect to the modal source involved. While in the deontic obligation interpretation, the modal source is identified with a human referent, it refers to circumstances or forces of nature in the practical necessity reading. How the modal source is instantiated in quantificational uses is less clear. As was illustrated in Section 2.2.1.3, raising verbs are characterised by three properties. First, lacking a subject argument, they can occur with a non-referential subject. Accordingly, in its practical necessity and in its obligation reading, müssen selects predicates that are subcategorised for non-referential subjects such as weather verbs and patterns that do not involve a subject at all: (108)

Es mussDEO (hier unbedingt noch) schneien. it must here absolutely still snow-inf ‘It is absolutely necessary that it snows here.’

(109)

... dass (unbedingt noch) getanzt werden mussDEO . that absolutely still dance-ppp pas.aux.inf must. ‘It is absolutely necessary that someone dance here.’

Note that müssen only exhibits a practical necessity interpretation in example (108). This is for pragmatic reasons, as it is rather unlikely to impose obligations on the weather, or to oblige someone to change the weather. In contrast, the example with impersonal passive (109), which does not involve any subject at all, can be interpreted as obligation without any problems. Deontic patterns of müssen, which do not carry a referential subject, can be easily found in corpora, as is indicated in examples (110)–(112): (110)

„Für jede einzelne Tat muss es eine Freiheitsstrafe geben”, sagt for each single crime must it a imprisonment give-inf says die Staatsanwältin in ihrem Plädoyer.⁷⁴ the attorney in her plea ‘ „For each single of these crimes, the accused must be sentenced with an imprisonment.” the attorney says in her plea.’

(111)

In Kanada muss künftig länger gearbeitet werden.⁷⁵ in Canada must henceforth longer work-ppp pass.aux-inf ‘In Canada, people will have to work longer in future.’

74 DeReKo: BRZ06/DEZ.00079 Braunschweiger Zeitung, 01/12/2006. 75 DeReKo: A12/MAR.14387 St. Galler Tagblatt, 31/03/2012.

2.2 Case studies |

(112)

79

Bleibt die Miete oder auch die Kaution aus, ist für den Vermieter stays the rent or also the deposit out is for the landlord umgehendes Handeln geboten. Jede unnötige Verzögerung immediate reaction required every unnecessary procrastination vertieft den möglichen Schaden. Mit Kündigung und increases the possible damage with cancellation and Klage muss nicht mehrere Monate gewartet werden.⁷⁶ complaint must neg more months wait-ppp pass.aux-inf ‘If the rent or the deposit is not paid, it is necessary for the landlord to react immediately. Every unnecessary procrastination can increase the damage. It is no longer obligatory to wait patiently a couple of months before considering cancelling the contract or even filing a complaint.’

Secondly, as has been pointed out by Stechow (2003: 203), Wurmbrand (1999: 606 2001: 192), raising verbs tolerate de dicto interpretations of quantified NPs. In the corpus examples (113) and (114) the indefinite NPs is interpreted in the scope of the modal necessity operator yielding a reading in which the NP does not refer to a particular individual. Whereas example (113) involves a practical necessity interpretation, a deontic interpretation turns out to be more plausible for example (114). (113)

Die Arbeit zu zweit ist auch deswegen unerlässlich, weil einer the work in two is also therefore indispensable because one von uns beiden immer wach sein muss, um die Piloten of us two always awake be-inf must in.order.to the pilots des Ballons zu leiten.⁷⁷ the-gen balloon to direct-inf ‘The work in pairs is also necessary for the particular reason that one of the two of us has to stay awake to pilot the aviators of the balloon.’

(114)

Wollen Jugendliche im Kubus eine Runde »töggelen«, erhalten want adolescent in.the cubus a round play receive sie einen Spielball. Bedingung: Einer muss das Handy als they a ball condition a must the cell.phone as Depot hinterlegen.⁷⁸ deposit deposit-inf ‘If teenagera want to play a round in the cubus, they will get a ball. There is a condition: one of them has to leave his mobile phone as a deposit.’

76 DeReKo: BRZ06/APR.00020 Braunschweiger Zeitung, 01/04/2006. 77 DeReKo: E99/MAR.06800 Zürcher Tagesanzeiger, 13/03/1999. 78 DeReKo: A10/JUN.03327 St. Galler Tagblatt, 10/06/2010.

80 | 2 Modal verbs: A class struggle

Likewise, example (115) clearly exhibits a deontic interpretation in a context in which a sergeant issues an order to his group of soldiers. In this particular context, he does not oblige a particular individual to guard the object. The quantifying expression ein needs to bear the accent in the NP in this type of configuration. Note that it is not relevant here whether ein and einer are used as an indefinite pronoun or as a numerical determiner in the examples given below, as Carpenter (1998: 87) has illustrated that numerical determiners behave like ordinary existential quantifiers.

Ein Mann muss die gesamte Nacht das Objekt bewachen. a man must the whole night the object guard-inf

(115)

‘During the whole night, one person has to guard the object.’

Thirdly, raising predicates are transparent with respect to voice. The sentences (116a) and (116b) involve infinitival complements that refer to the same state of affairs. They only differ with respect to how they encode their subjects: In the active example (116a), the subject is the underlying agent argument der Reinhold, in the passivised example the subject is identified with the underlying theme argument der Nanga Parbat. If müssen selected a proper subject argument, we would expect the obligation to be imposed on Reinhold in (116a), and on the Nanga Parbat in the passivised example (116b), yielding an unacceptable interpretation. But as it turns out, the two sentences do not differ with respect to their meaning. (116)

a.

Der Reinhold muß den Nanga Parbat ohne the Reinhold must the-acc Nanga Parbat without Sauerstoffgerät bezwingen. oxygen.apparatus conquer-inf ‘Reinhold has to conquer the Nanga Parbat without oxygen mask.’

b.

Der Nanga Parbat muß vom Reinhold ohne the-nom Nanga Parbat must by.the Reinhold without Sauerstoffgerät bezwungen werden. oxygen.apparatus conquer-ppp pass.aux-inf ‘The Nanga Parbat has to be conquered by Reinhold without oxygen mask.’

Both examples can be interpreted with either an obligation reading or a practical necessity interpretation. Essentially, müssen does not assign a semantic role to the subject NP. This becomes most obvious in the example that involves the passivised infinitive complement (116b). Being a mountain, the Nanga Parbat is not an appropriate target for bearing an obligation. Thus, müssen cannot identify its subject as the goal of the obligation in these examples. Corresponding configurations can easily be found in corpora, as is illustrated in (117):

2.2 Case studies |

(117)

81

Der Doktortitel muss aberkannt werden.⁷⁹ the doctor.title must deprive-ppp pass.aux-inf ‘The doctorate degree must be taken away.’

The example given above is a call based on laws and regulation. The only interpretation that is plausible here is a deontic one. Once again, müssen agrees with a subject that cannot be considered as its semantic argument. A doctorate degree is not a licit bearer of an obligation. Accordingly, the configuration illustrated above has to involve a raising pattern. As was pointed out in the previous section, Carlson (1977: 119) and Brennan (1993: 96) identify certain uses of the possibility modal can as a quantificational modal verb. In addition, Brennan briefly discusses some quantificational uses of the necessity modal verb will. However, it remains mysterious why the less marked necessity modal verb must cannot act as a quantifier over indefinite NPs. Unlike the English necessity modal verb must, its German counterpart müssen may occur in the scope of negation. In this type of configuration, it can occasionally exhibit an interpretation in which it acts as a quantifier over individuals. In example (118), nicht müssen serves as a negated universal quantifier over individuals (¬∀). It expresses that, in the set of good second-hand cars, there is at least one instance that is not worse than a new car. The remaining examples (119)–(123) behave in an according way. (118)

Ein guter Gebrauchtwagen muss nicht schlechter sein als ein a good second-hand.car must neg worse be-inf than a Neuwagen⁸⁰ new.car ‘A good second-hand car does not need to be worse than a new car.’

(119)

Vegetarisches Essen muss nicht langweilig sein.⁸¹ vegetarian food must neg boring be-inf ‘Vegetarian food does not need to be boring.’

(120)

Kunst muss nicht immer brotlos sein.⁸² art need neg always bread.less be-inf ‘Art does not need to be unprofitable.’

79 Der Spiegel 8/2011, p. 27, 21.02 2011. 80 DeReKo: NUZ11/MAI.01632 Nürnberger Zeitung, 18/05/2011. 81 DeReKo: BRZ11/JUN.06063 Braunschweiger Zeitung, 16/06/2011. 82 DeReKo: RHZ11/MAI.18218 Rhein-Zeitung, 16/05/2011.

82 | 2 Modal verbs: A class struggle

(121)

Ein Hund, der mit dem Schwanz wedelt, muss nicht unbedingt a dog that with the tail wags must neg necessarily freundlich gestimmt sein⁸³ cordially tempered be-inf ‘A dog that wags its tail is not necessarily friendly.’

(122)

Ein Haus ohne Salonbetrieb muss nicht seelenlos sein.⁸⁴ a house without salon.service must neg soul.less be-inf ‘A house without service in the salon does not need to be soulless.’

(123)

Der 4. Juli muss nicht immer ein deutscher Freudentag sein. Auf the 4 July must neg always a German joy.day be-inf at den Tag genau 44 Jahre nach dem ersten the day exactly 44 years after the first WM-Titelgewinn im Berner Wankdorfstadion (3:2 world.championship.title in.the Bernese Wankdorf.stadium (3:2 über Ungarn) hat Deutschland am Samstag in Lyon eine against Hungary) has Germany at.the Saturday in Lyon a der schmerzhaftesten Niederlagen hinnehmen müssen: im the-gen painful-sup defeat takemust in.the Viertelfinal an WM-Neuling Kroatien quarterfinal at world.championship.new.comer Croatia gescheitert, mit 0:3 verloren – klar und deutlich, ohne failed with 0:3 lost clearly and explicitly without Wenn und Aber.⁸⁵ if and but. ‘The 4th of July does not need to be always a German day of rejoicing. Exactly 44 years after the first victory in the world championship final at the Berner Wankdorfstadion, Germany had to accept a very painful defeat on Saturday in Lyon: In the quarterfinal of the World Championship, they clearly lost against the newcomer Croatia with a final score of 0:3’

As for the examples (118)–(122), a deontic interpretation does not come into consideration. It makes no sense to impose an obligation that a new car has to be better than a second-hand car, or that vegetarian food has to be boring. However, there are a lot of ambiguous examples, e.g. (123), which can be interpreted as the negation of a call, yielding a deontic interpretation. Similar instances are discussed by Welke (1965: 72) labelled as ‘variante 2’.

83 DeReKo: A09/NOV.00330 St. Galler Tagblatt, 02/11/2009. 84 DeReKo: A98/OKT.63556 St. Galler Tagblatt, 09/10/1998. 85 DeReKo: A98/JUL.45470 St. Galler Tagblatt, 06/07/1998.

2.2 Case studies |

83

It is worth mentioning that a subject NP in the scope of a negated universal quantifier is not always realised as an indefinite NP. In quite a lot cases, such NPs are represented by mass nouns, e.g. vegetarisches Essen ‘vegetarian food’ in example (119), and Kunst ‘art’ in (120). It remains to be shown how this quantification over the extension of a mass noun can be properly formalised. Finally, the question arises why these types of quantificational uses predominantly occur with the possibility modal verb können and the negated necessity modal verb müssen. This might be related to the position they occupy in the Aristotelian Square of Oppositions. Both the I and O corner host particularly valid propositions: The possibility modal verb können expresses a particular affirmative proposition, assigning a predication to some but not all items in its restrictor; the universal modal verb müssen expresses a particular negative proposition, assigning a predication to not all items in its restrictor. Thus, it seems that modal verbs in configurations that are related to a particular validity are more suitable to act as quantificational modals in German. Alternatively, the ¬□p could be interpreted as ♢¬p. Viewed from this point of view, the configuration could be seen as one which contains a possibility operator. Whether this phenomenon can be extended to other languages as well remains to be explored by future research. 2.2.2.4 Raising directionals with event modification Like können, the necessity modal verb müssen occurs fairly often along with verbless directional phrases. Once again, there are revealing instances of these configurations in which no infinitive can be inserted without leaving the interpretation of this utterance unaffected, as the discussion in the previous section has revealed. (124)

Jeans, Hemden, Jacken, Mäntel, Pullis, Anzüge, usw. – jeans shirts jackets coats sweater suits etc alles muss raus (#kommen/ #gebracht werden)!⁸⁶ everything must out get carried pass.aux-inf ‘Jeans, shirts, jackets, sweaters, suits, etc – everything must go!’

As has been already shown in Section 2.2.1.4, there are a lot of reasons to consider these usages of modal verbs with verbless directional phrases as independent patterns that cannot be accounted for in terms of an ellipsis of an infinitive.

86 DeReKo: A11/FEB.01302 St. Galler Tagblatt, 04/02/2011.

84 | 2 Modal verbs: A class struggle

2.2.2.5 Raising infinitives with clause modification As was already seen in the case of können, the epistemic uses of the necessity modal verb müssen are subject to the CoDeC: The proposition they embed cannot be part of the knowledge of the speaker (deictic centre). Moreover, they occur in environments in which circumstantial modal verbs are ruled out. They can modify predications that involve an identified individual and any predicate that refers to an event in the past (cf. 125 or 127), or a temporally unbound state (cf. 129 or 130). All of the epistemically modified utterances below imply that the proposition is not part of the speaker’s knowledge. (125)

Schreckliche Angst muss der kleine Junge gehabt haben, der terrible fear must the small boy have-ppp inf that am Samstag in ein tiefes Loch gefallen ist.⁸⁷ at Saturday in a deep hole fall-ppp is ‘The boy who fell into the deep hole on Saturday must have been terribly frightened.’

(126)

Ereignet haben müssen sich die Taten zwischen happen-ppp have-inf must refl the crimes between Donnerstag, 15 Uhr, und Dienstag, 6.30 Uhr.⁸⁸ thursday 15 o’clock and tuesday 6.30 o’clock ‘The crimes must have happened between Thursday 3 p.m. and Tuesday 6.30 a.m.’

(127)

Die Kleidungsstücke deuten dann auch darauf hin, dass es sich the clothes indicate then also to.it at that it refl um einen Mann gehandelt haben müsste.⁸⁹ about a man deal-ppp have-inf must-sbjv.pst ‘The clothes indicate that it must have been a man.’

(128)

Während sie in Mutters Tagebüchern gestöbert hatte, traf sie auf during she in mothers diaries rummage had met she on drei Männernamen: Sam, Bill und Harry. Sie lädt diese zu three male.names Sam Bill and Harry she invites them to ihrer Hochzeit ein. Einer von ihnen muss ihr Vater sein, der sie her wedding in one of them must her father be-inf that she zum Traualtar führen sollte.⁹⁰ to.the altar guide-inf should ‘While she was flipping through her mother’s diaries, she could find three male names: Sam, Bill and Harry. She invites them to her wedding. One of them must be her father who should guide her to the altar.’

87 88 89 90

DeReKo: BRZ09/JUL.16635 Braunschweiger Zeitung, 06/07/2009. DeReKo: BRZ09/APR.06547 Braunschweiger Zeitung, 16/04/2009. DeReKo: RHZ09/JUN.24827 Rhein-Zeitung, 29/06/2009. DeReKo: A09/JUL.00991 St. Galler Tagblatt, 03/07/2009.

2.2 Case studies |

85

Er muss der Täter sein, denn sein Sperma lässt sich in ihrem he must the culprit be-inf as his sperm lets refl in her Unterleib nachweisen.⁹¹ abdomen prove-inf

(129)

‘He must be the culprit because his sperm could be found in her pelvic area.’

Die Menschen schauen inzwischen William und Harry an und the people watch meanwhile William and Harry on and erkennen, dass Charles ein guter Vater sein muss⁹² recognise that Charles a good father be-inf must

(130)

‘Meanwhile, the people take a look at William and Harry and recognise that Charles must be a good father.’

Once again, the epistemic modal verb comes in two morphological realisations: As indicative muss, and as subjunctive of the past müsste. In contrast to the case of können, the use of indicative epistemic muss is significantly more frequent than its subjunctive counterpart. Even if the subjunctive is rare, it does exist, contradicting Lötscher (1991: 348), who claims that müsste can never be used as an epistemic modal verb. As in the case of epistemic könnte, the interplay of the subjunctive and the epistemic modal operator is rather complex and it is no trivial matter to unravel them. In some semantic respects müssen behaves analogously to its possibility denoting counterpart können. In a number of contexts, they are not interchangeable without affecting the interpretation. First of all, the indicative epistemic necessity modal verb muss occasionally occurs in the scope of negation. Like the epistemic possibility modal verb kann, these instances cannot be replaced with their subjunctive past counterpart müsste. (131)

a.

Der Sachverständige aus Koblenz machte der Sache ein the accident.assessor from Koblenz made the affair a Ende. Unter Berücksichtigung von Fahrverhalten und end under consideration of driving.behaviour and Bewegungsablauf, Schadensbild, Bodenbelag, path.of.motion damage road.surface Lichtverhältnissen sowie Geräuschkulisse kam er zu lighting.conditions as.well.as background.noise came he to

91 DeReKo: NUZ09/JUN.00298 Nürnberger Zeitung, 04/06/2009. 92 DeReKo: SOZ08/NOV.02694 Die Südostschweiz, 14/11/2008.

86 | 2 Modal verbs: A class struggle

dem Schluss, dass die Fahrerin den Unfall nicht the conclusion that the driver the accident neg bemerkt haben muss.⁹³ notice-ppp have-inf must ‘The accident assessor from Koblenz put an end to this affair. Considering the driving behaviour, the path of motion, the damage, the road surface, the lighting conditions and the background noise, he came to the conclusion that the driver does not need to have noticed the accident.’

b.

# [. . . ] kam er zu dem Schluss,

dass die Fahrerin den came he to the conclusion that the driver the Unfall nicht bemerkt haben müsste. accident neg notice-ppp have-inf must-sbjv.pst

Intended reading: ‘[. . . ] he came to the conclusion that the driver does not need to have noticed the accident.’

(132)

a.

Auch die Immobilienkrise in den USA muss noch nicht also the real.estate.crisis in the USA must still neg ausgestanden sein.⁹⁴ stand-ppp be-inf ‘It is not necessarily the case that the real estate crises in the US has already been overcome.’

b.

# Auch die Immobilienkrise in den USA müsste

noch also the real.estate.crisis in the USA must-sbjv.pst still nicht ausgestanden sein. neg stand-ppp be-inf

Intended reading: ‘It is not necessarily the case that the real estate crises in the US has already been overcome.’

In both examples that involve müsste (131b) and (132b), the epistemic necessity operator cannot be construed in the scope of the negation. This indicates that müssen with past subjunctive morphology is less acceptable in the scope of negation, if it is acceptable at all in such environments. Apart from that, epistemic subjunctive of the past müsste resembles the epistemic possibility modal verb könnte with subjunctive morphology in another crucial respect: It qualifies the underlying evidence in a similar way. Again, there seem to be different ways in which the subjunctive past morphology affects the meaning of epistemic müssen. First of all, there are cases in which müsste indicates that the evidence on which the epistemic conclusion is based involves nonverified premises:

93 DeReKo: RHZ06/FEB.12183 Rhein-Zeitung, 13/02/2006. 94 DeReKo: RHZ07/OKT.17666 Rhein-Zeitung, 19/10/2007.

2.2 Case studies |

(133)

87

Die Schuldfrage ist noch nicht gänzlich klar, doch nach ersten the guilt.question is still neg entirely clear but after first Ermittlungen der Bundespolizei scheint das Rotlicht an der investigation the-gen federal.police seems the red.light at the Bahnstrecke funktioniert zu haben. Demnach müsste der railroad.track work-inf to have-inf accordingly must-inf the Lastwagen-Fahrer bei Rot über die Gleise gefahren sein.⁹⁵ lorry.driver at red over the track drive-ppp be-inf ‘The issue of who is responsible is not entirely settled yet. According to the investigation carried out by the federal police, the traffic light seems to have worked. In correspondence, the driver of the lorry must have crossed the track when the light was red.’

(134)

Peter Westphal: „Wir haben die Oktobermieten gleich an Peter Westphal we have the October.rent immediately to Curanis/Vivacon überwiesen.” Demnach müsste Vivacon Curanis/Vivacon transfered accordingly must-sbjv.pst Vivacon die Mieten erhalten haben.⁹⁶ the rents receive-ppp have-inf ‘Peter Westphal: “We have immediately transfered the rent for October to Curanis/Vivacon” Correspondingly, Vivacon should have received the rent payments.’

(135)

Verursacher war wahrscheinlich ein Klein-Lkw oder Lkw, der bei causer was probably a small-lorry or lorry that at den Anhängerparkplätzen wendete und hierbei das Auto streifte. the trailer.parking turned and thereby the car scratch Es müsste sich um ein auffälliges it must-sbjv.pst refl about a noticeable Wendemanöver gehandelt haben, da auf der transposition.manoeuvre deal-ppp have-inf as at the Teerdecke entsprechender Reifenabrieb zu erkennen war.⁹⁷ road.surface corresponding tire.abrasion to recognise was ‘It was probably caused by a small lorry or a lorry that hit the car while turning at the trailer parking. It was obviously a noticeable transposition manoeuvre as the road surface exhibited corresponding traces of tire abrasion.’

(136)

Einer der Täter müsste sich die Kleidung während a the-gen culprit must-sbjv.pst refl the clothes during des Überfalls mit Blut beschmiert haben.⁹⁸ the robbery with blood stain-ppp have-inf

95 DeReKo: BRZ09/JUL.27453 Braunschweiger Zeitung, 31/07/2009. 96 DeReKo: BRZ08/JAN.13019 Braunschweiger Zeitung, 29/01/2008. 97 DeReKo: RHZ09/NOV.01570 Rhein-Zeitung, 03/11/2009.

88 | 2 Modal verbs: A class struggle

‘One of the culprits hypothetically has to have gotten blood all over his clothes during the robbery.’

A lot of the instances found in the corpus reveal the precise nature of the interplay between the epistemic modal verb and the subjunctive operator. Frequently, they explicitly refer to a premise that cannot be taken for granted. In example (133), the adverb demnach is linked to the first shaky results of the investigation carried out by the police indicating that the traffic light was properly functioning. In a similar fashion, the adverb demnach refers to the statement made by Peter Westphal in example (134). Though not explicitly, müsste in example (135) refers to the prior assumption that a lorry probably caused the damage while it was turning. Apart from that, there are also examples of epistemic müsste that do not involve reference to a premise that has been made explicit in the prior discourse, such as example (136). As it seems, the prior discourse includes a non-verified presupposition that describes the circumstances of the robbery, such as that one of the culprits was so close to the victim that a contact with blood was inevitable. In such configurations, the speaker signals critical distance with respect to the validity of these premises. But crucially, these premises in the examples above are not counterfactual or known to be false. The speaker is just not entirely convinced. In contrast to könnte, müsste can be replaced with its indicative cognate muss more easily in these instances. There appear to be subtle changes in the interpretation, as will be shown below in more detail. Secondly, there are interpretations of epistemic müsste that are based on counterfactual premises. Analogous examples have been provided by Mortelmans (2000: 206). (137)

Guido Niedermann fand am Waldboden eine Feder. »Ganz Guido Niedermann found at forest.ground a feather very deutlich ist zu sehen, dass diese Feder abgebissen wurde, clearly is to see that this feather off.bite-ppp was folglich war dieses Federvieh Opfer eines Marders oder thus was that poultry victim a-gen marten-gen or Fuchses. Wäre die Feder ausgerupft worden, fox-gen. be-sbjv.pst the feather pinch-ppp pass.aux.ppp müsste der Täter ein Greifvogel gewesen sein«, erklärte must-sbjv.pst the culprit a raptor be-ppp be-inf said Niedermann.⁹⁹ Niedermann

98 DeReKo: RHZ06/NOV.27737 Rhein-Zeitung, 28/11/2006.

2.2 Case studies |

89

‘Guido Niedermann found a feather in the forest. “It can be seen very clearly that this feather was bitten off. Thus, this poultry was a victim of a marten or a fox. If the feather were plucked, it would follow that the culprit must have been a bird of prey. ” said Niedermann.’

(138)

Wenn alle Meldungen über Schwangerschaften der if all reports about pregnancies the-gen Oscar-Preisträgerin gestimmt hätten, müsste sie oscar-winner attune-ppp have-sbjv.pst must-sbjv.pst she mittlerweile 30 Babys bekommen haben. Kidman ist Mutter meanwhile 30 babies get-ppp have-inf. Kidman is mother zweier adoptierter Kinder.¹⁰⁰ two-gen adopted-gen children ‘If all of those reports about the Oscar winner’s pregnancies had been true, then she would have had 30 babies by now. Kidman is the mother of two adopted children.’

In both cases, it is evident that the premise expressed by the antecedent of the conditional is counterfactual: In example (137), the speaker makes it explicit that the feather was not plucked. Likewise, the person who utters the sentence in (138) refutes the claim that all reports that Kidman was pregnant were true. If müsste is replaced with muss in the environments in which the epistemic conclusion is based on a counterfactual premise, an important difference becomes apparent with respect to the non-verified premise readings. Although müsste can be replaced with muss in non-verified premise readings, such replacement yields a diminished degree of acceptability in the counterfactual premise readings. The subjunctive morpheme of müsste is necessary to identify the premise as a counterfactual one. This clearly indicates that the interaction of the subjunctive operator and the epistemic operator can be construed in terms of semantic composition in counterfactual premise interpretations. This further illustrates that there are good reasons to distinguish between the non-verified and the counterfactual premise interpretation of müsste. While the latter is an evident result of a compositional interplay of the counterfactual operator and the epistemic operator, the status of the first type is less clear. As wwas shown in the preceding section, there are two ways of expressing the subjunctive of the past in German: with the synthetic form (müsste), and with an analytic form consisting of the subjunctive past auxiliary würde and a bare infinitive (müssen). If the counterfactual premise interpretation is indeed the result of semantic composition, it is expected that it should also be available if the subjunctive operator würde and epistemic modal operator (müssen) are combined in 99 DeReKo: A00/FEB.13497 St. Galler Tagblatt, 22/02/2000. 100 DeReKo: BRZ07/DEZ.11819 Braunschweiger Zeitung, 31/12/2007.

90 | 2 Modal verbs: A class struggle

a periphrastic manner, and indeed such instances of the counterfactual premise interpretation can be found: (139)

Der Autor: „Die Zukunft gehört der Tätigkeitsgesellschaft, in the author the future belongs the occupation.society in der Erwerb und Nichterwerb, Arbeit und Leben eine Einheit which earning and non-earning, work and life a union bilden und das Zeitdenken Vorrang vor dem constitute and the time.thinking priority before the Gelddenken hat. ”Demnach würde uns freie Zeit in money.thinking has accordingly sbjv.pst.aux us free time in Zukunft wichtiger sein müssen als Gut und Geld. future important-comp be-inf must-inf than goods and money Seit 1990 registrieren Statistiker aber einen gegenläufigen since 1990 register statistician yet a contrary Trend.¹⁰¹ trend ‘The author claims: “The future will be dedicated to the occupation society in which earning and non-earning, work and life constitute a union and time based reasoning prevails money based reasoning.” Accordingly, it would be the case that spare time must be more important to us in future than money and goods. However, since 1990 statisticians have been observing the opposite trend.’

In example (139), the speaker refers to an author’s hypothesis about the future of our society that he considers to be false. It appears to be much more difficult, if it is possible at all, to find an appropriate context in which a periphrastic form of the subjunctive of the past of müsste can be construed with a non-verified premise interpretation. Though not obvious, a compositional analysis of the non-verified premise interpretation does not seem to be entirely excluded. Yet, it remains to be shown how these readings could be derived in a compositional way. At this point the question arises of what the precise nature of these counterfactual operators is with respect to the status of the epistemic conclusion. It seems to be clear that the counterfactual premise interpretations occur predominantly in counterfactual (irrealis) conditionals. Accordingly, one could conclude that those cases in which epistemic müsste is not modified by a wenn-clause should be considered as truncated counterfactual (irrealis) conditionals. According to Kasper (1987: 24–28), the semantic contribution of the subjunctive past operator is that the modified proposition cannot be felicitously uttered.

101 DeReKo: NUN98/MAI.01774 Nürnberger Nachrichten, 23/05/1998.

2.2 Case studies |

91

When this concept is applied to epistemic modal verbs, this would indicate that the epistemic conclusion is not drawn in the actual world, as one of the premises on which it is grounded is known to be false. More precisely, the speaker would signal that he is not in a position to draw this conclusion in the actual world, and that he would be able to do so in a minimally different possible world. Rephrasing the epistemic necessity modal operator in possible world semantics, this would yield the following circumscription: In a minimally different possible world, it is consistent with all of the possible worlds that are consistent with the speaker’s knowledge in that hypothetical, minimally different world. As seems to become clear, the description of epistemic modal operators in the scope of a counterfactual operator requires possible worlds that are interpreted in another, non-actual, minimally different world. This is a very intricate matter that cannot entirely be solved here. Likewise, it needs to be investigated to what extent a hypothetical assumption involves an assumption in the real world. Even if a speaker using müssen might indicate that he is not in a position to draw this particular conclusion, he nevertheless communicates an epistemic evaluation of the modified proposition. At least at some meta-level, he makes a claim about the validity of the proposition in the actual world. In a similar fashion, this reasoning could equally apply to the non-verified premise interpretation of müsste. In this type of environment, the speaker would signal that he does not know whether the premise holds or not. If he knew that the premise were true, then he would draw the epistemic conclusion expressed by the epistemic modal verb. An analysis that considers the entire act of conclusion as counterfactual could also account for those cases in which the speaker knows the embedded proposition to be false. As was pointed out in the preceding section, the usage of an epistemic modal operator indicates that the embedded proposition is not part of the speaker’s (deictic centre’s) knowledge. In the most canonical case, the speaker would not know that the embedded proposition is false either. This is most natural for counterfactual readings, as in examples (137) or (138). It has yet to be shown to what extent such configurations, where the modified proposition is known to be false, are also compatible with a non-verified premise interpretation. Until this point, epistemic müsste behaves very similarly to epistemic könnte. But whereas kann can be easily replaced with its subjunctive counterpart könnte in most environments without drastically affecting the overall interpretation, muss cannot be replaced with müsste a number of contexts:

92 | 2 Modal verbs: A class struggle

(140)

Der Wagen rollte vom Bahnsteig auf das Gleis und wird the car rolled from.the platform on the rail.track and is prompt von einem einfahrenden Zug erfasst und 40 Meter promptly by a approaching train hit and 40 Meter mitgeschleift. Der Bub muss einen besonderen Schutzengel with.dragged the boy must a special guardian.angel gehabt haben: Er kam mit leichten Verletzungen davon.¹⁰² have-ppp have-inf he came with minor injuries away ‘The car rolled from the platform onto the rail track and it was immediately hit by a passing train and dragged along a distance of 40 meters. The boy must have had a special guardian angel: he survived with minor injuries.’

(141)

Es muss ein schlimmes Bild gewesen sein, das sich der it must a bad picture be-ppp be-inf that refl the Frau bot, die gestern morgen früh als erste am woman offered that yesterday morning early as first at.the Unfallort eintraf.¹⁰³ accident.location arrived ‘For the woman, who arrived first at the accident yesterday early in the morning, it must have been a terrible scene to see.’

(142)

Eine Henauerin hat erzählt, dass sie vor dem Kirchgang a Henauerian has told that she before the church.going dem Bäcker einen Apfel gebracht habe. Nach dem the baker an apple brought have-sbjv.prs after the Kirchgang hat sie ihn wieder abgeholt, schön im Teig church.going have she him again up.picked beautifully in dough gebacken, und es muss so gut geschmeckt haben!¹⁰⁴ baked and it must so good taste-ppp have-inf ‘A woman from Henau said that before she went to church, she brought the baker an apple. After church, she picked it up again, baked in dough, and it must have tasted so good!’

(143)

Dabei muss er das Fahrzeug einer 29 Jahre alten Frau thereby must he the vehicle a-gen 29 year old woman übersehen haben, die mit ihren Kindern auf dem Weg zu miss-ppp have-inf who with her children on the way to einem Freizeitpark war.¹⁰⁵ a amusement.park was

102 DeReKo: A09/OKT.04501 St. Galler Tagblatt, 17/10/2009. 103 DeReKo: A09/JUL.07445 St. Galler Tagblatt, 30/07/2009. 104 DeReKo: A09/NOV.01850 St. Galler Tagblatt, 06/11/2009.

2.2 Case studies |

93

‘In doing so, he must have missed the car of a 29 year old woman who was driving her children to an amusement park.’

(144)

Der Anblick, der sich den Rettern bot, muss schrecklich the sight that refl the rescuer offered must awful gewesen sein.¹⁰⁶ be-ppp be-inf ‘It must have been such an awful sight for the rescuers.’

(145)

Rund um mein Hotel in der Innenstadt gibt es nur around around my hotel in the city.centre gives it only kostenpflichtige Parkplätze. Das muss ich wohl übersehen with.costs parking that must I obviously miss-ppp haben – prompt klebte an meinem Mietwagen ein Ticket.¹⁰⁷ have-inf promptly sticked at my hire.car a ticket ‘Around my hotel in the city centre, there are only parking spaces where you have to pay to park. I must have missed that and there was soon a parking ticket stuck to my car.’

(146)

Es muss eine Herkulesarbeit gewesen sein, das weitversprengte it must a hercules.work be-ppp be-inf the widely.scattered Notenmaterial der 1813 uraufgeführten »Medea in Corinto« zu sheet.music the-gen 1813 premiered Medea in Corinto to sammeln¹⁰⁸ collect-inf ‘It must have been a Herculean task to collect the widely scattered sheet music of the “Medea in Corinto”, which premiered in 1813.’

In all of the examples (140)–(146), a substitution of muss with müsste would affect the interpretation. Interestingly, the epistemic necessity modal muss can be more successfully be replaced with the epistemic probability modal verb dürfte in the given instances. This indicates that müsste cannot be considered as a necessity modal that involves a necessity that is slightly weaker than the one of muss; otherwise it would be expected that the replacement by dürfte should lead to an even lesser degree of acceptability. The analysis provided by Fritz (1997: 101) and Mortelmans (2000: 205), who argue that the presence of a subjunctive of the past morpheme weakens the degree of necessity, can therefore not be correct. These examples constitute a revealing case for the identification of the precise semantics of muss and müsste. After müsste has been inserted in an example 105 106 107 108

DeReKo: BRZ09/JUL.25816 Braunschweiger Zeitung, 27/07/2009. DeReKo: A09/FEB.01731 St. Galler Tagblatt, 07/02/2009. DeReKo: BRZ09/JAN.04683 Braunschweiger Zeitung, 12/01/2009. DeReKo: A09/OKT.04815 St. Galler Tagblatt, 19/10/2009.

94 | 2 Modal verbs: A class struggle

as the ones given in examples (140)–(146), it just needs to be analysed how the context needs to be changed in order to make müsste fully acceptable again, as is illustrated in the modified version of example (140): (147)

Der Bub müsste einen besonderen Schutzengel gehabt the boy must a special guardian.angel have-ppp haben have-inf ‘??(In this case) The boy must have had a special guardian angel.’

In order to repair the context according to example (147), one of the underlying premises needs to be adjusted. In the version with the indicative epistemic modal verb muss (cf. 140), the speaker bases his conclusion on at least two verified premises: {the boy has been dragged along by a train, the boy has survived with minor injuries }. In order to create an appropriate context for müsste, one of these verified premises has to be refuted or labelled as ‘non-verified’ or ‘counterfactual’. Depending on the context, it appears that the use of müsste requires the most prominent premise to be non-verified or counterfactual. This could, again, be a hint that the instances of müsste with subjunctive past involve a conditional configuration in which the non-verified premise is hosted by the antecedent. In some cases, the antecedent is explicitly realised as a wennclause; in the other cases, it remains implicit. This also explains why epistemic müsste cannot express conclusions that are based on direct evidence from the utterance situation. In this case, the most prominent premise would be a state of affairs that is accessible by all sorts of senses; therefore, the premise would be verified already and impossible to contradict. This behaviour reveals the nature of the subjunctive morpheme on müsste: It acts as a qualifier for the underlying evidence involved. The evidence upon which the epistemic conclusion is drawn needs to include premises that are not verified or even known to be false. Concluding, epistemic müssen comes in two guises: The indicative form muss, and the subjunctive of the past müsste. As in the case of epistemic könnte, epistemic müsste occurs in two functions: It either indicates that one of the premises on which the epistemic conclusion is based is not verified, or it indicates that one of the premise is counterfactual. Unlike its indicative counterpart, the subjunctive epistemic necessity modal verb müsste is not acceptable in the scope of negation. In all of these aspects, müsste strongly resembles the subjunctive epistemic possibility modal verb könnte. But these verbs differ with respect to the extent to which they are interchangeable with their indicative cognates: Whereas the subjunctive epistemic necessity modal verb müsste can be more easily replaced with the indicative muss, the subjunctive epistemic possibility modal verb könnte can-

2.2 Case studies |

95

not be substituted with kann in most contexts without affecting the interpretation. In contrast, the indicative epistemic modal necessity verb muss withstands substitution with its subjunctive counterpart müsste, whereas the indicative epistemic possibility verb kann can always be substituted with its subjunctive cognate könnte.

2.2.3 wollen The volitional modal verb wollen is well studied and there is considerable consensus about its behaviour. The semantic range covers the expression of volition and reported speech, as well as volition that originates in a referent distinct from the subject referent. Volitional semantics have remained fairly stable through the course of history, as shown by Bech (1951) and Fritz (1997: 44). Following the tradition established by Bech (1949: 38), wollen is generally considered as necessity modal verb that involves a modal source that lies within the subject referent. 2.2.3.1 Transitive uses As has been noticed on various occasions, there are instances of wollen that occur with an accusative NP without infinitive, as has been pointed out by Raynaud (1977: 5, 20). The status of these occurrences is contested. On the one hand, there are authors such as Öhlschläger (1989: 69), who argue that these instances involve an ellipsis of the infinitive. On the other hand, there are authors such as Zifonun (1997: 1255), Erb (2001: 96) and Eisenberg (2004: 97), who argue that in these uses, wollen serves as a transitive verb. These authors illustrate their claim with passivised examples of wollen. Such patterns are also documented in corpora, as shown in examples (148)–(150): (148)

‘Man müsse zur Kenntnis nehmen, daß der Einzug von one must at notice take that the move-in-nom of Ausländern in Gemeindewohnungen einfach von den Leuten nicht foreigners in flats.of.the.township simply by the people neg gewollt wird, meinte Häupl.¹⁰⁹ want-ppp pass.aux said Häupl ‘One has to acknowledge that the people simply do not want that foreigners move into township’s flats.’ lit: The move in is not wanted by the people

109 DeReKo: N93/FEB.06544 Salzburger Nachrichten, 22/02/1993.

96 | 2 Modal verbs: A class struggle

(149)

Nur der Waffenstillstand hat Bestand der von only the ceasefire has continuance that-rel.prn.nom from innen her kommt und von allen Kriegsparteien gewollt inside par comes and by all war.parties want-ppp wird¹¹⁰ pass.aux ‘A ceasefire can only be succesful if it is proposed from the inside and if it is wanted by all of the warring parties.’ lit: A ceasefire is wanted by all of the warring parties

(150)

Die politisch-planerisch Verantwortlichen müssen begreifen, the political-planner-adj responsible.person must understand daß vor allem ein Theaterzentrum mit sozio-kultureller that foremost a theater center with socio-cultural Bedeutung von den Menschen dieser Stadt gewollt wird¹¹¹ relevance by the people of.this city want-ppp pass.aux ‘The people responsible for political planning must understand that the people of this city want a theatre centre with socio-cultural relevance.’ lit: That a theatre centre is wanted

Öhlschläger (1989: 69) acknowledges that wollen can occasionally be passivised, but from his perspective, it is not fully productive. As a consequence, wollen with an accusative NP has to involve an ellipsis of an infinitive. According to Öhlschläger (1989: 69), it is always the infinitive haben ‘have’ that is elided here. Yet, Öhlschläger’s arguments are not plausible for two reasons. First of all, passivised wollen occurs much more frequently than passivised können. This is remarkable, as word forms related to können are much more frequent than those related to wollen in the archive W of the DeReKo corpus, upon which the study is based. So if können is considered as “fully productive” with respect to its passive, it remains mysterious why the passive of wollen should be “less productive”. Whoever considers können to be a transitive verb has to consider wollen, as well. Secondly, if Öhlschläger (1989) were right in his assumption that wollen without an infinitive complement should be considered as an ellipsis of haben, it is expected that the examples in (148)–(150) should be derived from configurations in which the infinitive haben is spelled out overtly. However, after inserting an infinitive of haben, the acceptability of such configurations drastically decreases, as is highlighted in examples (151)–(153).

110 DeReKo: P91/NOV.08961 Die Presse, 29/11/1991. 111 DeReKo: R98/DEZ.102438 Frankfurter Rundschau, 19/12/1998.

2.2 Case studies |

(151)

97

* Der Einzug von Ausländern in Gemeindewohnungen wird

the move-in of foreigners in flats.of.the.township pass.aux von den Leuten nicht haben gewollt by the peopleneg have-inf want-ppp (152)

* Ein Waffenstillstand wird

von allen Kriegsparteien a ceasefire pass.aux.pst by all war.parties haben gewollt have-inf want-ppp

(153)

* Ein Theaterzentrum wird

a theater.center gewollt want-ppp

von den Menschen haben pass.aux.pst by the people have-inf

The contrast between examples (148) and (151) remains unaccounted for in an analysis that treats wollen with an accusative NP as an ellipsis of the infinitive haben. There are further instances of wollen with NP that cannot be complemented with the infinitive haben, e.g. the question pattern in (154a): (154)

a.

A: Was willst du hier (*haben)? what want you here have-inf

b.

B: Ich will mit dir reden. I want with you talk-inf A: ‘What do you want here?’ B: ‘I want to talk to you’

The wh-pronoun does not seem to be a canonical VP-anaphor. If this was the case, it should be possible to substitute wollen in example (154a) with other verbs such as dürfen or müssen, but this is not possible. Nevertheless, it seems to be possible to coordinate an accusative NP complement with a bare infinitive complement of wollen in some contexts. (155)

Wir wollen Sonne statt Reagan, ohne Rüstung leben.¹¹² we want sun instead.of Reagan without weapons live-inf ‘We want sunshine rather than Reagan, to live without arms.’

As is widely assumed, coordination requires the identity of the categories of both constituents. A more detailed discussion is given by Dougherty (1970: 850, 864), Jackendoff (1977: 51), Gazdar (1981: 157, 173), Schachter (1984: 269) and Pollard and Sag (1994: 202). Thus, the pattern in example (155) would presuppose that

112 Joseph Beuys and BAP, election spot for the Green Party released in April 1982.

98 | 2 Modal verbs: A class struggle

Sonne ‘sun’ has the same category as the infinitive complement ohne Rüstung leben. Accordingly, the constituent Sonne must be part of an infinitive that has an elided verbal head. Even if an analysis in terms of ellipsis might be attractive for this example, there are strong reasons for the existence of instances of transitive wollen, as the data on passivisation indicates. Alternatively, the pattern in (155) could be considered asymmetric coordination. 2.2.3.2 dass-Sätze As has been pointed out by Becker (1836: 181), Welke (1965: 78), Raynaud (1977: 6, 20), Fritz (1997: 17), Erb (2001: 96) and Eisenberg (2004: 96) wollen occasionally selects finite dass-clauses. Welke (1965: 78) suggests that the dass-clause is the result of a transformation that has been applied to the infinitive complement. Likewise, Öhlschläger (1989: 70) stresses that, in these patterns, the dass-clause fulfils the same function as the infinitive complement. Therefore, he argues that these configurations do not involve an ellipsis of an infinitive. (156)

Unser Chef will nicht, dass so viele Ausländer bei uns drin our boss wants neg that so many foreigners in us in sind¹¹³ are ‘Our boss does not want so many foreigners enter our place.’

(157)

Nur 32 Prozent wollten, dass Schwarz-Gelb weitermacht.¹¹⁴ only 32 percent wanted that black-yellow continues ‘Only 38 percent wanted that the Black-Yellow coalition stays in power.’

This clearly indicates that the infinitive is not mandatory with wollen. The fact that the subject of wollen and the subject of the dass-clause are disjoined in reference in the most canonical case has motivated some authors, such as SzumlakowskiMorodo (2006: 325), to assume that this configuration is due to a syntactic restriction. In contrast, Vater (2001) and Reis (2001: 303) have illustrated that the two subjects can occasionally bear the same index. Such patterns are also found in corpora: (158)

Ich will, dass ich meinen Kindern auch Elefanten im I want that I my children also elephants in.the Tiergarten zeigen kann.¹¹⁵ zoo show-inf can

113 DeReKo: BRZ07/JUN.09793 Braunschweiger Zeitung, 12/06/2007. 114 DeReKo: HMP11/MAR.02571 Hamburger Morgenpost, 28/03/2011.

2.2 Case studies |

99

‘I also want to be able to show my children the elephants in the zoo.’

However, configurations in which the subject referent of the matrix clause and the subject referent are identical are undoubtedly rare and rather exceptional. There seems to be yet another restriction on subjects referent. The anti-performative restriction observed by Hinterwimmer (2014) seems to hold for wollen, too. Unlike an imperative (236), wollen with dass-clause cannot be used to directly refer to an addressee out of the blue (160). The latter pattern asserts the existence of an intentional act settled in the common ground. (159)

Geh jetzt! go-imp now ‘Go now!’

(160)

# Ich will, dass Du jetzt gehst!

I

want that you now go

‘I want you to go now!’

But then, example (160) becomes felicitous if the speaker already uttered an imperative like (236) and uses wollen with dass-clause to insist on his wish. 2.2.3.3 Control infinitives with event modification The status of volitional wollen with an infinitive is less controversial. Most authors, such as Bech (1949: 5), Welke (1965: 78), Raynaud (1977: 19), Höhle (1978: 84) and Diewald (1999: 140), assume that it involves a proper referential subject argument that encodes the modal source besides its infinitive argument. In more recent approaches, these verbs are considered as control verbs. Accordingly, Öhlschläger (1989: 119), Kiss (1995: 162), Reis (2001: 302), Axel (2001: 40), Erb (2001: 78), Wurmbrand (2001: 170) and Abraham (2001: 18, 2002: 36, 2005: 241, 257, 261) classify the volitional use of wollen as a control verb. It is not uncontroversial that it assigns a semantic role to its syntactic subject. (161)

Zu Guttenberg will das Jagdbombergeschwader 33 Zu Guttenberg wants the fighter.bomber.squadron 33 modernisieren.¹¹⁶ modernise-inf ‘Zu Guttenberg wants to modernise the 33rd fighter bomber squadron.’

Becker (1836: 181) considers wollen a necessity modal verb with a particular specification. Likewise, Bech (1949: 5, 39) argues that the strength of the modal re-

115 DeReKo: NUN11/MAR.02889 Nürnberger Nachrichten, 28/03/2011. 116 DeReKo: RHZ10/JAN.06017 Rhein-Zeitung, 15/01/2010.

100 | 2 Modal verbs: A class struggle

lation expressed by wollen resembles the one expressed by müssen and sollen, rather than the one encoded by können, dürfen and mögen. In a similar way, Calbert (1975: 36 Fn.2), Ehrich (2001: 165) and Remberger (2010: 165, 169) point out that wollen is most appropriately analysed as a necessity modal verb that involves a necessity which has its source within the subject argument referent. It can be distinguished from the control use of müssen, which expresses a physical need by means of the ordering source. The universal quantifier contributed by wollen quantifies over those worlds that are consistent with the preferences of the speaker. 2.2.3.4 Control directionals with event modification Like können and müssen, the volitional verb wollen can frequently be found with verbless directional phrases. (162)

Deutschland will weg vom Atomstrom.¹¹⁷ Germany wants away from nuclear.electricity ‘Germany wants to move away from nuclear electricity.’

As was demonstrated in Section 2.2.1.4, configurations such as the one in (162) cannot be efficiently captured as ellipses of an infinitive. 2.2.3.5 Raising infinitives with event modification Bech (1949: 9) has already acknowledged that wollen occasionally occurs with inanimate subject NPs. As the syntactic subject is not a licit bearer for an experiencer role, a canonical volitional interpretation is not appropriate. In a similar fashion, Welke (1965: 85) observes uses of wollen with impersonal subjects which involve a semantics which cannot easily be captured. In later works such as those by Stechow and Sternefeld (1988: 446), Öhlschläger (1989: 170), Kiss (1995: 161, 167), Reis (2001: 302) and Wurmbrand (2001: 170), these occurrences are considered as the raising use of wollen. As some of these authors emphasize, wollen does not appear with a figurative meaning here. The inanimate subject does not receive an anthropomorphic interpretation resulting in a metaphoric volitional reading. There is no semantic relation at all between its syntactic subject and wollen in these cases. This is most convincingly illustrated by the fact that this use of wollen occurs with non-referential subjects. It seems that at least three different types of wollen with raising structure have to be differentiated. First of all, there is one type which preferably occurs in negative polarity environments (cf. 163–168). Brinkmann (1962: 367) already observed 117 DeReKo: HMP11/MAR.01535 Hamburger Morgenpost, 17/03/2011.

2.2 Case studies |

101

that there is a use of wollen which occurs only in negative contexts, and which does not encode volition of the subject referent. Instead, it expresses that the event denoted by the embedded infinitive continues to happen against the expectation of the discourse participants. (163)

Heute will es scheinbar gar nicht mehr aufhören zu regnen.¹¹⁸ today want it obviously intn neg more stop-inf to rain-inf ‘Today, it does not seem to want to stop raining.’

(164)

Als das Filmteam um »die Knef« in Nürnberg ankam, wollte As the film.crew around the Knef in Nürnberg arrive wanted es nicht aufhören zu regnen.¹¹⁹ it neg stop-inf to rain-inf ‘When the film crew, along with ‘the Knef’, arrived in Nuremberg, it did not want to stop raining.’

(165)

Auch nach 24 Jahren Wahnsinn und wahnsinnigen 17 Studioalben also after 24 years madness and madly 17 studio will es einfach nicht ruhig werden um die want it simply neg become quiet around the Thrash-Metal-Supernova „MegaDave” Mustaine.¹²⁰ Trash-Metal-Supernova “MegaDave” Mustaine ‘Even after 24 years of madness and an unbelievable 17 studio albums, there is nothing that can stop the Trash-Metal-Supernova “MegaDave” Mustaine.’

(166)

Ein Interessent wartet schon lange auf den erlösenden Anruf an interested waits already long for the liberating call Haiders, aber es will einfach nicht läuten: Der F-Sprecher Haider-gen but it want simply neg ring: The F-spokesman im Landesschulrat, Erich Petschacher.¹²¹ in.the regional.education.authority Erich Petschacher ‘There is a person who is interested, who has been waiting for Haider’s liberating call for a long time, but it just won’t ring: the F spokesman in the regional education board: The spokesman of the F in the regional education authority, Erich Petschacher.’

(167)

Irgendwie will und will es nicht Frühling werden.¹²² somehow want and want it neg spring become ‘Somehow, it does not want to be spring.’

118 DeReKo: RHZ02/FEB.07812 Rhein-Zeitung, 12/02/2002. 119 DeReKo: NUN04/MAI.01869 Nürnberger Nachrichten, 21/05/2004. 120 DeReKo: HMP07/JUN.01431 Hamburger Morgenpost, 14/06/2007. 121 DeReKo: K99/JUN.45893 Kleine Zeitung, 22/06/1999. 122 DeReKo: RHZ09/MAR.09602 Rhein-Zeitung, 11/03/2009.

102 | 2 Modal verbs: A class struggle

(168)

Im Gegensatz zu seinen Klassenkameraden, will und will bei in contrast to his class.mates wants and wants at ihm kein einziges Schamhaar wachsen, obwohl er doch wie him no single pubic.hair grow although he par like toll in Katharina verschossen ist.¹²³ insane in Katharina mad is ‘In contrast to his classmates, not a single pubic hair to grow, even though he has a crush on Katharina.’

(169)

MAX: [. . . ] Das will mir nicht gefallen.¹²⁴ this wants me neg please ‘This does not please me.’

Example (169) illustrates that this interpretation is already documented for the late 18th century. Typically, this variant of negated wollen co-occurs with the reinforcing sentence adverb einfach ‘simply’, as illustrated in examples (165)–(166) or re-duplication of wollen (cf. 167–168). It merits closer attention that the string will und will seems to be restricted to negative polarity environments: among 239 hits found in the DeReKo corpus based on the query with the string will und will, there is not a single one without a negation. This is a strong indicator in favour of an analysis that treats this use of wollen as negative polarity item.¹²⁵ The exact interpretation of the negative polar raising verb wollen is hard to capture. Even if the examples (163)–(168) do not involve an overt bearer of the volition, it appears that such a referent is contextually required. All of the examples describe a state of affairs that does not happen to be. But moreover, they are only fully acceptable if there is a referent who wants that state of affair to happen. The utterance in example (167) is only felicitous if there is a referent who wants it to be spring. Obviously, this contribution is most efficiently captured by the presupposition ‘x wants p to happen’. The interpretation of the negative polar raising verb wollen is, to some extent, reminiscent of the behaviour of concessive epistemic mögen, as has been pointed out by Welke (1965: 110), Allard (1975: 69, 70), Öhlschläger (1989: 187) and Diewald (1999: 236). Both patterns involve some referent who is the source of a volition, but that is not overtly encoded in the clause. This analysis is furthermore indirectly supported by Schoetensack (1856: 294). As he observes, there are further uses of mögen that involve a volition attributed to a third party, as in er möge hereinkom-

123 DeReKo: M05/MAI.35869 Mannheimer Morgen, 02/05/2005. 124 Friedrich von Schiller Wallensteins Tod, II. Aufzug, 7. Auftritt (1799). 125 Corpus query carried out on 21th March 2012 based on the string will ’’und’’ will.

2.2 Case studies |

103

men ‘He likes-sbjv.prs enter / fig. He may enter’. In this case, möge expresses that the embedded proposition is consistent with the wishes of the speaker. A contrasting analysis is suggested by Gergel and Hartmann (2009). They assume that wollen, in its volitional interpretation, is a raising verb which does not give up its ability to assign a thematic role to identify the source of volition. In the canonical case, it is the subject argument of the embedded infinitive that raises to the syntactic subject position of wollen (SpecVP) and receives the experiencer role from wollen. Furthermore, they argue that depending on the type of embedded predicate the raised argument can also be a dative object. In the case of impersonal verbs such as gelingen ‘succeed’ or schmecken ‘taste’, the least oblique argument is an experiencer dative NP which is claimed to be raised into the subject of wollen, where it should be identified as source of volition, according to the analysis put forth by Gergel and Hartmann (2009: 337). (170)

Ihm will einfach nichts gelingen.¹²⁶ him-dat wants simply nothing succeed ‘He just doesn’t succeed at anything (although he tries....)’

(171)

Dem Großvater will die Suppe nicht schmecken.¹²⁷ the-dat grandfather wants the soup-nom neg like ‘The grandfather does not like/want to like the soup.’

According to their analysis, the experiencer argument of wollen is identified with the dative NP ihm in example (170), and the dative NP dem Großvater in example (171). As a consequence, they conclude that all modal verbs in German are raising verbs (General Raising Hypothesis). Some of them, such as wollen, assign thematic roles to their syntactic subjects nevertheless. Thus, Gergel and Hartmann (2009: 350) assume that raising into theta position is possible, abandoning the classical Theta Criterion. However, their account suffers from a major shortcoming. There are examples which are very similar to the one they discuss and which do not contain any NP that could potentially be identified as the source of volition, or the experiencer argument of wollen, e.g. cases in which the embedded predicate is klappen ‘work out’, as in example (172). (172)

Die Jungs haben brutal gekämpft, aber es will einfach nicht the boys have brutally struggled but it want simply neg klappen mit der Goldmedaille.¹²⁸ work.out with the gold.medal

126 As quoted in Gergel and Hartmann (2009: 331). 127 As quoted in Gergel and Hartmann (2009: 331).

104 | 2 Modal verbs: A class struggle

‘The boys fought hard but it just won’t be enough for the gold medal.

Lacking an appropriate animate argument, klappen does not provide an appropriate candidate that could be identified as the source of volition. Moreover, it involves a non-referential subject, which in turn indicates that it can only be embedded by a raising verb. Similar reasoning applies to the examples (163)–(168), they cannot be accounted for by the analysis suggested by Gergel and Hartmann (2009). The alternative outlined here is to analyse the examples (170) and (171) discussed by Gergel and Hartmann (2009) not as instances of oblique raising of dative objects into theta positions, but to subsume them under a more general phenomenon. Accordingly, they could be analysed along the same lines as the ones above. Furthermore, these examples are not instances of ‘weak wollen’ discussed by Ehrich (2001: 165). According to Bech (1949: 5), she assumes that some instances of wollen carry an underlying possibility operator, rather than a necessity operator. In any case, as these uses also need to overtly specify their source of volition, they could not account for the patterns in (163)–(168) and (172), as these clauses do not involve appropriate arguments. Apart from the negative polar raising uses of wollen, there seem to be at least two further types of raising patterns. As has been shown by Helbig and Buscha (2001: 121), there are occurrences of wollen which seem to express a mere necessity, and which can be replaced with müssen without affecting the interpretation too much. Arguably, these uses should also be considered as raising verbs. Such uses become evident with embedded predicates that are passivised. Often, they involve a promoted theme argument that is inanimate. As has been shown in the Sections 2.2.1.3 and 2.2.2.3, such environments are typical of raising verbs. As can be seen, this use of wollen is occasionally modified by the adverbs erst and einmal. (173)

Der Name des Coiffeurgeschäfts will deshalb gut the name the-gen hairdresser.shop-gen wants therefore well gewählt sein.¹²⁹ choose-ppp be-inf ‘Therefore, the name of the hair saloon needs to be well chosen.’

128 DeReKo: M11/MAR.01983 Mannheimer Morgen, 05/03/2011. 129 DeReKo: A09/MAI.04520 St. Galler Tagblatt, 14/05/2009.

2.2 Case studies |

(174)

105

Dazu kommen die Länder Osteuropas, deren to.it come the countries Eastern.Europe-gen rel.prn.gen Sprung in die Marktwirtschaft auch erst einmal finanziert leap in the market.economy also first once finance-ppp sein will.¹³⁰ be-inf wants ‘In addition to this, there are the Eastern European countries, whose leap into a market economy also needs to be financed.’

(175)

Ach, du schöne Ferienzeit. Und was wäre sie ohne oh you beautiful holiday and what was she without Vorfreude? Doch will die schönste Zeit des Jahres anticipation but wants the beautiful-sup time the-gen year-gen auch gut vorbereitet sein.¹³¹ also well prepare-ppp be-inf ‘Oh, you beautiful holidays! And what would it be without pleasant anticipation? But the most wonderful time of the year also needs to be well prepared.’

(176)

Eine Lok ohne Wagen kostet schon einige Euro. Und a locomotive without cars cost already some Euro and auch eine gute Werkzeugausstattung will erst einmal bezahlt also a good tool.equipment wants first once pay-ppp sein.¹³² be-inf ‘A locomotive without cars already a lot of money. And also, a good tool kit needs to be paid for first.’

(177)

Die Betreiber des Capitol schätzen den entstandenen the operators the-gen Capitol-gen estimate the caused Schaden auf rund 60 000 Euro – ein Loch, das erst einmal damage at about 60 000 Euro a hole that first once gestopft werden will.¹³³ plug-ppp pass.aux wants ‘The operators of the Capitol estimate that the damage amounts to 60 000 Euros – a hole that one has yet to close/that is not so easy to close.’

130 DeReKo: NUN90/OKT.01381 Nürnberger Nachrichten, 20/10/1990. 131 DeReKo: BRZ09/APR.00836 Braunschweiger Zeitung, 02/04/2009. 132 DeReKo: RHZ07/DEZ.22127 Rhein-Zeitung, 27/12/2007. 133 DeReKo: M02/SEP.68278 Mannheimer Morgen, 13/09/2002.

106 | 2 Modal verbs: A class struggle

(178)

Auch hier wird sich erweisen, daß [. . . ] nicht nur einfach also here will refl prove that neg only simply Glanz zu ernten ist, sondern Jahr für Jahr zwölf Monate harte brilliance to harvest is but year for year twelve months hard organisatorische Arbeit den Erfolg begründet. Diese Arbeit will organisational work the success bases this work wants erst einmal geleistet sein.¹³⁴ first once perform-ppp be-inf ‘Here, too, it will be shown, that [. . . ] not just brilliance is to be harvested, but that the basis of this success is hard organisational work, 12 months a year, year after year. This work has yet to be done.’

The precise interpretations of these uses of wollen are difficult to capture. In example (173), wollen certainly does not express a volition that is attributed to the subject der Name des Coiffeurgeschäfts ‘the name of the hair saloon’. Rather, this use does not encode any volition at all. Without significantly altering the interpretation of the clause, it can be substituted with the necessity modal verb müssen. Yet there are some subtle semantic differences that remain to be captured. Finally, there are obvious raising uses of wollen which contribute a meaning that is even more difficult to isolate. These cases appear to occur with verbs like scheinen ‘seem’. (179)

Fast will es scheinen, als müsse „Derevo” an diesem almost wants it appear-inf as must Derevo on this Freitagabend hoch über der Stadt das Ende seines Stücks friday.night high over the city the end it-gen piece-gen alleine feiern.¹³⁵ alone celebrate-inf ‘It seems almost as if ‘Derevo’ will have to celebrate the end of his piece alone high above the city.’

(180)

Der See wollte heute randlos erscheinen.¹³⁶ the lake wanted today borderless appear-inf ‘The lake appeared to go on forever.’

As has been shown, wollen occurs in a couple of environments as a raising verb. However, in each of the three cases discussed above, the semantic contribution of wollen is rather hard to capture.

134 DeReKo: P94/SEP.30244 Die Presse, 09/09/1994. 135 DeReKo: RHZ03/JUL.21748 Rhein-Zeitung, 29/07/2003. 136 DeReKo: WAM/EFP.00000 Walser, Martin: Ein fliehendes Pferd. – Frankfurt a.M., (1978), p. 126.

2.2 Case studies |

107

2.2.3.6 Raising directionals with event modification The negative polar raising use also occurs with verbless directional phrases. The examples (181)–(183) involve inanimate subject referents that are not likely to be identified as a source of volition. Moreover, they share some preferences with the negative polar raising pattern of wollen. First of all, wollen expresses a similar meaning. Secondly, it frequently occurs with the reinforcing adverb einfach ‘simply’ (cf. 182) and occasionally, it can be found with a re-duplication of the finite verb (cf. 183). (181)

inzwischen ist es ein Uhr nachts, aber mir will die meanwhile is it one o’clock night but me wants the Geschichte nicht aus dem Sinn.¹³⁷ story neg out the mind ‘Meanwhile, it is one o’clock in the morning, but I don’t want to get the story out of my mind.’

(182)

Ich möchte lernen, aber der Schulstoff will einfach nicht in I want learn-inf but the lesson want simply neg in meinen Kopf!¹³⁸ my head ‘I would like to revise, but I simply can’t keep the material in my mind.’

(183)

Stäheli zittert, bangt, verzweifelt – der Puck will und will Stäheli shivers trembles despairs the puck wants and wants nicht ins Tor.¹³⁹ neg in.the goal ¨ shivers, trembles, and despairs – but the puck just doesn’t want to go into the ‘Stheli goal.’

(184)

Das will mir nicht ein.¹⁴⁰ this wants me neg in ‘I cannot understand this.’

As with the raising use of wollen, which embeds infinitives, this pattern is attested for the 18th century, as illustrated in example (184).

137 Fyodor Mikhaylowich Dostoyewsky, Der Idiot, translated by Svetlana Geier, p. 221. 138 DeReKo: A00/NOV.78588 St. Galler Tagblatt, 15/11/2000. 139 DeReKo: A08/MAR.06383 St. Galler Tagblatt, 15/03/2008. 140 Gotthold Ephraim Lessing, Nathan der Weise, IV. Aufzug, 7. Auftritt, (1779).

108 | 2 Modal verbs: A class struggle

2.2.3.7 Control infinitives with clause modification As was already observed by Becker (1836: 181), Schoetensack (1856: 294), Curme (1922: 322) and Bech (1949: 6), in some instances wollen refers to a claim attributed to the subject referent. Some authors, such as Öhlschläger (1989: 233), Abraham: 11 (2001: 11, 2002: 27, 2005) and Reis (2001: 287 Fn. 1), assume that this use of wollen is identified as the epistemic reading of wollen. Indeed, this use of wollen shares a couple of the essential characteristics with canonical epistemic modal verbs such as können and müssen. As soon as wollen refers to a claim of the subject referent, it can embed a predication consisting of an identified individual and a predicate that refers to an event in the past (cf. 185– 187), or a predicate that denotes permanent states that cannot be changed (cf. 188– 189). This parallel behaviour has already been pointed out by Abraham (2001: 11, 2005). (185)

Sieben Packerl Rotwein will er vor dem Prozess seven packets red.wine wants he before the process konsumiert haben.¹⁴¹ consume-ppp have-inf ‘He claims to have consumed seven boxes of red wine prior to the trial.’

(186)

So will sie eine „Depression mit psychotischer Färbung” bei So wants she a depression with psychotic color at Pleger erkannt haben.¹⁴² Pleger recognise-ppp have ‘With Pleger/ In Pleger, she claims to have seen “depression with a touch of psychotic behaviour”.’

(187)

Sein Landsmann Frederick Albert Cook will bereits am 21. April his countryman Frederick Albert Cook wants already at 21 april 1908 dort gewesen sein.¹⁴³ 1908 there be-ppp be-inf ‘His fellow countryman Frederick Albert Cook claims to have already been there on 21 April 1908.’

(188)

Er will angeblich der alleinige Täter sein.¹⁴⁴ he wants reportedly the sole culprit be-inf ‘He claims to be the sole culprit.’

141 142 143 144

DeReKo: NON09/JUL.08001 Niederösterreichische Nachrichten, 15/07/2009. DeReKo: NON09/JUL.02654 Niederösterreichische Nachrichten, 07/07/2009. DeReKo: RHZ06/NOV.30695 Rhein-Zeitung, 30/11/2006. DeReKo: HMP09/NOV.00548 Hamburger Morgenpost, 05/11/2009.

2.2 Case studies |

(189)

109

Badhapur ist ein Sadhu, ein Weiser, Gerechter. 106 Jahre will die Badhapur is a Sadhu, a sage righteous 106 years wants the hagere Gestalt mit dem langen grauen Haar schon alt rawboned figure with the long grey hair already old sein.¹⁴⁵ be-inf ‘Badhapur is a Sadhu, a wise man, a religious man. This haggard form with long grey hair already claims to be 106 years old.’

Note that the canonical circumstantial volitional interpretation is not possible in the examples given above. In early descriptions such as Schoetensack (1856: 294), reportative wollen was analysed as a pattern that involves a kind of ellipsis. As he argues in more detail, there is a mediating clause that has been elided: er will, (dass man glaube), dass er ihn gesehen habe ‘He wants (that one thinks) that he has seen him’. As Bech (1949: 6) emphasises, the canonical volitional use of wollen targets the “realisation” of the embedded predication. The uses above, however, target the “reality” of the embedded predication. This contrast corresponds exactly to the one between circumstantial interpretations of können or müssen and their epistemic counterparts. As Truckenbrodt (2006: 263–268) amongst others, has pointed out, any sentential speech act type conveys a volitional component. As with assertions, the speaker wants the addressee to add the embedded proposition to the Common Ground. As it seems, the function of the reportative use of wollen is to express this volitional component of declarative clauses and associate it with the subject referent. In opposition to canonical epistemic modal verbs, the use of wollen discussed above is not subject to the CoDeC in the same way. Some authors, such as Palmer (1986: 72), Schenner (2009) and Faller (2010: 661), argue that it merits a different name: ‘quotative’ or ‘ reportative’ modal verb. Whereas an epistemic modal verb indicates that the modified proposition is not part of the speaker’s knowledge, the proposition that is embedded reportative wollen (cf. 185–189) can, in principle, be part of the speaker’s knowledge, as will be shown in more detail in Chapter 5. As will be seen there, the relevant attitude holder for the evaluation of reportative wollen seems to be the subject referent, rather than the speaker. In contrast to epistemic modal verbs, the speaker may agree or disagree with the modified proposition which is labelled as a claim of another referent. He may even know that it is true or false. Similar observations have been made by Öhlschläger (1989: 235), Ehrich (2001: 157), Colomo (2011: 241) and Faller (2011: 4, 2012: 289).

145 DeReKo: NUN99/OKT.02110 Nürnberger Nachrichten, 23/10/1999.

110 | 2 Modal verbs: A class struggle

As has been shown in Sections 2.2.1.5 and 2.2.2.5, epistemic modal verbs can be characterised in terms of the environments from which they are excluded. Reis (2001: 294, 296) observes that reportative wollen occurs more readily in environments in which epistemic modals are difficult to interpret or entirely excluded, such as non-finite environments or in questions. Thus, there are two aspects in which reportative wollen differs from canonical epistemic modal verbs. It is clear that these two aspects might be derived from a major syntactic difference. Whereas epistemic modal verbs are always raising verbs which are evaluated with respect to a super-ordinate attitude holder, which is the speaker in most cases, wollen remains to be a control verb, even in its quasiepistemic use, as has already been stressed by Öhlschläger (1989: 121) and Reis (2001: 302). The crucial difference is that reportative wollen introduces the attitude holder as its proper argument. Assuming that each epistemic modal operator contains a variable for the epistemic agent who undertakes the evaluation, the variable for the epistemic agent is always locally bound by the subject referent in the case of reportative wollen. Being already instantiated, the variable contributed by the operator is not subject to the strict conditions of identifications anymore. Accordingly, the reportative modal verb can be used more flexibly. The precise mechanism of identification will be discussed in Chapter 5. Finally, reportative wollen turns out to be an ideal candidate to check the nature of wh-clefts. Thráinsson and Vikner (1995: 62) and Erb (2001: 88) assume that in wh-clefts, control verbs can only be separated from their infinitive complements, whereas raising verbs cannot. According to them, the subject in the whclause needs to be licensed by an external theta role. This would explain why epistemic modal verbs are ruled out in such configurations, as, being raising verbs, they lack an external theta role. If their view is correct, wollen could be expected to be separated from its infinitive complement in wh-clefts, as is illustrated in (190), which is derived from the corpus example (186): (190) ?? Was sie vielmehr will, ist bei Pleger eine „Depression mit

what she rather wants is at Pleger a depression with psychotischer Färbung” erkannt (zu) haben.¹⁴⁶ psychotic color recognise-ppp to have-inf Intended reading: ‘What she claims is to have recognized that Pleger has “depression with a touch of psychoticism.” ’

In this example, the licencing conditions are fulfilled. The finite verb in the whclause assigns a theta role to the subject NP sie. Nevertheless, the examples that involve reportative instances of wollen seem to be less acceptable than those cases with volitional wollen, which are discussed in Section 4.8. If Thráinsson

2.2 Case studies |

111

and Vikner (1995: 62) and Erb (2001: 88) are right in claiming that only control verbs can be separated from their infinitive complements, the availability of an external theta role cannot be the sole condition here. As will be shown in Section 4.8, there are alternative explanations. In some rare cases, the referent to which the claim is attributed can be an argument different from the matrix subject. In example (191), reportative wollen embeds the idiomatic pattern ein Begriff sein with a dative NP that encodes the experiencer argument. Crucially, it is the dative NP that bears the most prominent thematic role in this pattern, according to the hierarchy of thematic roles proposed by Dowty (1991). Surprisingly, reportative wollen can embed such a type of predicates in which the experiencer is identified with a dative NP, rather than a nominative NP. Nevertheless, the referent to which the claim is attributed is the dative NP Strasser rather than the matrix subject Mensdorff-Pouillys Firma Valurex. (191)

Auch Mensdorff-Pouillys Firma Valurex, die also Mensdorff-Pouilly-gen enterprise-nom Valurex-nom rel.prn in dem Deal als Drehscheibe fungierte, will Strasser bis in the deal as turning.device functioned wants Strasser-dat until heute kein Begriff sein („Valurhops oder wie die today no-nom notion-nom be-inf valurhops or how that heißt”).¹⁴⁷ called.is ‘Strasser claims that Mensdorff-Pouilly’s enterprise ‘Valurex’, which functioned as the key turning point in the deal, was not known to him („Valurhops, or whatever it is called”).’

This example indicates that the instance of reportative wollen cannot be a canonical control verb as the semantic role is evidently not assigned to the matrix subject but to a dative object that depends on the embedded predicate ein Begriff sein. Moreover, an interpretation as a non-reportative instance of the raising pattern of wollen is not plausible in the example above, in which an accused former minister claims that he was not informed about a certain deal. Yet, it has to be checked what repercussion this example has on the analysis of reportative wollen. It is conceivable that this pattern is not generally accepted. It could turn out that this configuration could be accounted for by the oblique raising analysis illustrated by Gergel and Hartmann (2009: 337). As they assume, wollen can discharge its semantic role that is designated to the subject argument, alternatively, to the dative argument of the embedded infinitive. But as has been illustrated in Section 2.2.3.5, their analysis has some shortcomings that have yet to be overcome.

147 http://www.orf.at/stories/2126743/2126744/, accessed on 20th June 2012.

112 | 2 Modal verbs: A class struggle

Note that this example also involves a negative quantifier in subject position. As will be shown in Section 2.2.8.7, such configurations exhibit a mysterious interaction with reportative control verbs. 2.2.3.8 Raising infinitives with clause modification Likewise, there are instances of clause modifying wollen that do not seem to select a referential subject argument. In these uses, wollen is highly reminiscent of concessive epistemic mögen, which is discussed in more detail in Section 2.2.7.7. As is typical of sentences that involve a modal operator with concessive meaning, the main clauses in examples (193) and (194) are followed by a clause that is introduced by the adversative conjunction aber ‘but’. (192)

Warum hat Schwenker, wenn es tatsächlich so gewesen sein why has Schwenker if it indeed so be-ppp be-inf wollte, das nicht gleich erzählt?¹⁴⁸ wanted that neg immediately told ‘If it really was like that, why didn’t Schwenker say that immediately?’

(193)

Der etwa 69 800 Mark teure Cross Country [. . . ] will the about Mark expensive cross country wants zwar kein Geländewagen sein, doch in seiner Nähe although no all-terain.vehicle be but in his closeness sehen ihn die Volvo-Leute schon.¹⁴⁹ see him the Volvo-people yet ‘Though the Cross County, which costs about 69,800 DM, may not be an off-road vehicle, it is considered by the Volvo people as something comparable.’

(194)

Ich will den Arbeitgebern sicher nicht gefallen wollen, I want the employer certainly neg please-inf want-inf aber ich möchte in schwierigen Zeiten ein Optimum für die but I would.like.to in difficult periods an optimum for the Arbeitenden herausholen.¹⁵⁰ employees get.out-inf ‘Certainly, I might not please the employers but I would like to get the maximum for the employees in such difficult times such as now.’

Some speakers prefer to analyse the inanimate subject referent in example (193) as an anthropomorphic subject. This analysis would not account for the case where

148 DeReKo: HMP09/DEZ.00650 Hamburger Morgenpost, 07/12/2009. 149 DeReKo: M97/712.03322 Mannheimer Morgen, 10/12/1997. 150 DeReKo: E98/NOV.28535 Züricher Tagesanzeiger, 07/11/1998.

2.2 Case studies |

113

wollen selects a non-referential subject in example (192), neither for example (194), in which wollen does not only occur as a matrix verb but also as the infinitive complement. The only plausible interpretation is one that is similar to the one of concessive-epistemic mögen. The speaker who utters such a configuration indicates that he acknowledges that the embedded proposition is possible, even if he considers the content of this proposition irrelevant to the ongoing discourse. At this point, it is important to emphasise that both verbs, wollen and mögen, express a volitional meaning in control structures. It seems, that parts of these semantic features are still active in concessive epistemic use. Thus, there are good reasons to acknowledge that wollen has an independent use as an epistemic raising verb, even if this use is rather marginal. This reasoning is further supported by instances of other volitional verbs such as sollen, which also exhibit a concessive resonance, as in example (265), discussed in Section 2.2.6.4.

2.2.4 dürfen The case of dürfen ‘be.allowed.to’ turns out to be of particular interest. As will be demonstrated, its indicative use can never be used with an epistemic interpretation, in contrast to all other items that are considered as modal verbs. However, with a subjunctive of the past morphology, an epistemic reading becomes available. As can be clearly seen, it is not plausible to subsume the epistemic use of dürfte under the use of dürfen: Whereas circumstantial dürfen involves a modal force that corresponds to a possibility, epistemic dürfte cannot be considered as a possibility verb anymore; rather, it is a verb that expresses probability. As most authors, such as Kratzer (1991: 650), implicitly assume that the modal force for a lexical item always remains the same, dürfen and dürfte have to be considered as two separate independent lexical items. In contrast to the previous cases, dürfen does not involve transitive uses in Contemporary German, although some such uses are attested in the Early New High German period, in which dürfen occurs without an infinitive complement. But at this point, it is not entirely clear whether these uses are due to confusion with the verb bedürfen ‘need’. The most important uses of dürfen are the permissive uses with bare infinitive complements. As already illustrated in Section 2.1.4, dürfen does not occur very frequently. 2.2.4.1 Transitive Uses There are a couple of cases occurring in texts from the Early New High German period in which dürfen is only combined with a NP. Interestingly, there seems to

114 | 2 Modal verbs: A class struggle

be a certain flexibility with respect to case assignment. Examples (197)–(200) are taken from the Deutsches Wörterbuch (DWB). As the editor of the relevant entry argues, dürfen can either select genitive NPs, as in (199)–(200), or accusative NPs, as in (197)–(198). This alternation merits closer attention, as the four examples all come from the same text. (195)

Der Oberst meinte/ es were ein betrug dahinder/ vnd sagte/ the colonel thought it was a deception behind and said Wahr der Sie solten hinfahren/ denn er duerffte they should away.go because he may-sbjv.pst the-gen good nicht¹⁵¹ neg ‘The colonel thought it was a deception and said that they should leave because he did not need their goods.’

(196)

Aber die buecher seines Gesetzes hette er nicht mit sich but the books his-gen law-gen had he neg with him gefuehrt/ denn dieweil er auff dem wasser were/ duerffte carried because while he on the water was may-sbjv.pst er jr nicht.¹⁵² he they-gen neg ‘But he did not carry along the books of his laws because while he was on water he did not need them.’

(197)

von art seind sie verderbt, geneigt zu bosheit darumb dörfen of art are they foul inclined to evil therefore may sie freund die sie underweisen, ermanen und strafen und they friends rel.prn them instruct warn and punish and von den sunden helfen¹⁵³ from the sin help ‘Their character is foul and inclined to evil, therefore they need friends that instruct, warn, punish and help them to refrain from sinning.’

(198)

da darf man wenig salz zů einer schüssel vol¹⁵⁴ there may one little salt for a bowl full ‘Little salt is needed for a entire bowl.’

151 152 153 154

Ulrich Schmid Neuwe Welt, 32a, (1567). Ulrich Schmid Neuwe Welt, 9a, (1567). Geiler von Keisersberg, sünden des munds, 47a (1518), as cited in the DWB. Geiler von Keisersberg, sünden des munds, 53a (1518), as cited in the DWB.

2.2 Case studies |

(199)

115

es darf wenig uszlegens, jederman weiszt waz es it may little interpret-inf.noun-gen everyone knows what it ist¹⁵⁵ is ‘Not much interpretation is required as everyone knows what it is’

(200)

wir dörften wol einer leiteren, sollen wir anders im nach we may well a-gen ladder-gen shall we further him after steigen¹⁵⁶ step-inf ‘We need a ladder if we would like to follow him.’

But crucially, these instances of dürfen without an infinitive complement express a meaning that is very different from the one created by dürfen with an infinitive complement. The transitive examples denote a need and thus reflect the original meaning of dürfen, as has been discussed by Fritz (1997: 10) in some detail. In contrast, dürfen with an infinitive is negative polar at this period and expresses a permission. As these two meanings essentially differ with respect to their modal force, it is not plausible to assume that they can be subsumed under the same lexicon entry. Note that there is an etymologically related verb bedürfen ‘need’ that selects a genitive NP. It is fairly likely that all of the examples discussed above emerged due to a contamination with bedürfen + genitive NP. This would explain the availability of the genitive case in these instances. As for those complements in the occurrences above that are considered as accusative NPs (cf. 197–199), it is not so clear whether they indeed carry accusative case. The example (197) is in principle ambiguous, and (199) involves a quantifier that selects a genitive NP. There is a further argument that speaks against the assumption that the examples above are archaic remnants of the transitive use of dürfen. During the Middle High German period, dürfen is not very frequent and predominantly found as a negative polar item with an infinitive complement, as has been illustrated by Bech (1951: 14). In the voluminous novels from the late 12th century, Pârzival and Iwein, dürfen can only be found with an infinitive complement and negation. Note that not all of the patterns illustrated above involve genuine transitive configurations. Upon closer inspection, it turns out that dürfen was also possible as an impersonal verb that lacked a referential subject argument (199).

155 Geiler von Keisersberg sünden des munds, 23a (1518), as cited in the DWB. 156 Geiler von Keisersberg sünden des munds, 86a (1518), as cited in the DWB.

116 | 2 Modal verbs: A class struggle

However, there are transitive uses of dürfen that frequently occur in processes of first language acquisition. In contrast to the examples given above, dürfen in (201) refers to a permission. (201)

Darf ich ein Eis? may I a ice.cream ‘Am I allowed to get ice cream?’

As these uses do not belong to the grammar of the target language, they will not receive any further attention here. Even if the examples discussed here cannot be considered as genuine transitive uses of the verb dürfen, such uses exist in earlier stages, as Birkmann (1987: 161) shows. 2.2.4.2 Raising infinitives with event modification The most frequent use of dürfen is the one in which it selects a bare infinitive complement. In most of these cases, it denotes permission. Following the tradition established by Bech (1949: 18), authors such as Welke (1965: 105) and Öhlschläger (1989: 162) assume that permission is most efficiently defined in terms of volition. In more detail, deontic dürfen indicates that the embedded proposition is consistent with the wishes of another referent, the person who grants the permission. As was indicated in Section 2.2.1.3, the precise syntactic status of circumstantial modal verbs with infinitive complements is contested. Exponents of the analysis put forth by Ross (1969: 86) assume that all circumstantial modal verbs are uniformly control predicates. Likewise, Welke (1965: 107) argues that deontic dürfen is a two-place predicate. In contrast, the alternative view is based on the assumption that circumstantial modal verbs can be raising verbs as well. As Öhlschläger (1989: 105) argues, the permissive uses of dürfen involve a raising structure. Applying the diagnostics for raising introduced in Section 2.2.1.3, it turns out that dürfen is indeed documented in environments that are only compatible with raising verbs. First of all, it can select non-referential subjects (cf. 202–205). Apart from that, the possibility operator contained in dürfen can take scope over existentially quantifying subject NPs yielding a de dicto-interpretation (cf. 206–207), which, according to Stechow (2003: 203), indicates raising. In both examples, the permission is not tied to an previously identified referent. (202)

Es darf nicht nur ums Sparen gehen.¹⁵⁷ it is.allowed.to neg about saving go-inf ‘It should not just be about saving money.’

157 DeReKo: HAZ09/FEB.01718 Hannoversche Allgemeine, 10/02/2009.

2.2 Case studies |

(203)

117

Nun ist aber alles wieder im Reinen, und es darf now is but everything again in.the pure and it is.allowed.to gefeiert werden.¹⁵⁸ celebrate-ppp pass.aux-inf ‘Now that everything has been sorted out, we are again allowed to celebrate.’

(204)

In Deutschland wohnt die Freiheit. Hier darf geraucht in Germany lives the freedom here may smoke-ppp werden.¹⁵⁹ pass.aux-inf ‘Freedom lives in Germany. Here, smoking is allowed.’

(205)

Es darf kein Schweigen mehr geben – Schweigen it is.allowed.to no silence more give-inf silence bedeutet Mitschuld¹⁶⁰ means complicity ‘There should not be silence any more – silence means complicity’

(206)

Einer seiner Sprösslinge darf den Kuchen a his-gen chip.off.the.old.block-gen is.allowed.to the cake schneiden, der andere sein Stück aussuchen.¹⁶¹ cut-inf the other his piece choose-inf ‘One of his chips off the old block may cut the cake, the other one may choose his piece.’

(207)

Dabei soll jede Band versuchen, mit maximal drei Titel thereby shall each band try with maximally three titles die Jury zu überzeugen. Einer der Songs darf the jury to convince-inf a the-gen song-gen is.allowed.to jedoch noch nicht auf einem existierenden Tonträger sein.¹⁶² but yet on a existing sound.carrier be-inf ‘In doing so, each band should try to win over the jury presenting no more than three titles. One of the songs must have not yet been released on an existing form of audio media.’

In both examples (206) and (207), the subject NP could be replaced with a NP that contains a canonical existential quantifier: ein Sprössling or ein Song. It is not important here whether ein is used as an indefinite pronoun or as a numerical

159 DeReKo: RHZ07/JAN.10458 Rhein-Zeitung, 13.01.2007. 159 DeReKo: BRZ06/AUG.12221 Braunschweiger Zeitung, 25/08/2006. 160 DeReKo: A00/MAI.36865 St. Galler Tagblatt, 27/05/2000. 161 DeReKo: RHZ02/MAR.16434 Rhein-Zeitung, 23/03/2002. 162 DeReKo: O94/FEB.15123 Neue Kronen-Zeitung, 19/02/1994.

118 | 2 Modal verbs: A class struggle

determiner, as Carpenter (1998: 87) has demonstrated that a numerical determiner behaves like an ordinary existential quantifier. Finally, dürfen in its permission reading is transparent with respect to voice. If the subject position of dürfen was assigned a semantic role and identified as the bearer of the permission, the examples given in (208a) and (208b) would be expected to refer to a distinct state of affairs. In the first case, the permission would be granted to Reinhold and in the second case, to the mountain Nanga Parbat. However, this is not the correct interpretation of these examples. (208)

a.

Der Reinhold darf den Nanga Parbat ohne the Reinhold is.allowed.to the-acc Nanga Parbat without Sauerstoffgerät bezwingen. oxygen.apparatus conquer-inf ‘It is allowed that Reinhold conquers the Nanga Parbat without oxygen mask.’

b.

Der Nanga Parbat darf vom Reinhold ohne the-nom Nanga Parbat is.allowed.to by.the Reinhold without Sauerstoffgerät bezwungen werden. oxygen.apparatus conquer-ppp pass.aux-inf ‘It is allowed that the Nanga Parbat is conquered by Reinhold without oxygen mask.’

Having shown that the permissive use of dürfen involves a raising pattern, it is now time to take a closer look at the semantic content. How can it be captured? As Kratzer (1981: 40) suggests, all of the traditional modal verbs in German can be considered items that involve quantification over possible worlds. Kratzer (1991: 649) further argues that the interpretation of any modal expression in natural language can be captured by means of three dimensions: modal force, modal base and ordering source. The dimension that is the least difficult to identify is the modal force. Bech (1949: 18, 38) uses almost the same semantic description for the permissive use of dürfen as for the possibility modal verb können. Furthermore, Becker (1836: 178) Kratzer (1981: 46) and Öhlschläger (1989: 158, 162) explicitly analyse deontic darf as a possibility modal verb. However, none of these authors provide sound empirical evidence for their conclusions. Nevertheless, there are good reasons to adopt their analyses, which treat dürfen as some sort of possibility modal verb. First of all, können occasionally exhibits a permissive interpretation which is obviously synonymous with dürfen, as was indicated in Section 2.2.1.3. Moreover, Levinson (2000: 36) has pointed out that quantifying expressions such as possibility modal adverbs induce scalar im-

2.2 Case studies |

119

plicatures.¹⁶³ If a speaker utters that something is possible, this utterance will imply that it is not certain. (209)

Possibly, there’s life on Mars. +> not certainly

From this, in turn, it follows that ((♢p) & (♢¬p)) should always be true in natural spoken language. Accordingly, it is expected that the conjunction of dürfen (p) and dürfen (¬p) should not result in a contradiction if dürfen indeed carries a possibility modal operator. As it turns out, deontic dürfen is acceptable in such configurations without causing a contradiction, irrespective of whether it is inflected for indicative (cf. 210a) or subjunctive of the past (cf. 210b). The diacritic || indicates a intonation break and the underlined constituent bears a high pitch accent. (210)

a.

Sie darf den Anruf entgegennehmen aber sie she is.allowed.to the call answer-inf but she darf ihn genau so gut auch || nicht entgegennehmen. is.allow.to him exactly as well also neg answer-inf ‘She is allowed to answer the call, but at the same time she is also allowed not to answer it.’

b.

In diesem Falle, dürfte sie den Anruf entgegennehmen in this case is.allowed.to she the call answer-inf aber sie dürfte ihn genau so gut auch || nicht but she is.allow.to him exactly as well also neg entgegennehmen. answer-inf ‘In this case, she would be allowed to answer the call, but at the same time she would also be allowed not to answer it.’

As these contrasts indicate, circumstantial dürfen indeed turns out to be a true possibility modal verb, confirming the views held by Bech (1949: 18, 38) and Kratzer (1981: 46). As has been observed by Öhlschläger (1989: 186) and Diewald (1999: 232), the indicative of dürfen lacks an epistemic interpretation. This type of interpretation can only be rendered with the subjunctive of the past form dürfte. Nevertheless, there are examples of indicative dürfen in which the speaker appears to express an epistemic evaluation with respect to the truth of the embedded proposition, see examples (211)–(213):

163 At this point, I would like to thank Roland Schäfer and Uli Reich for inspiring comments on this diagnostic.

120 | 2 Modal verbs: A class struggle

(211)

Das darf doch nicht wahr sein, dass der letzte grössere this is.allowed.to yet neg true be-inf that the last big Laden im Dorf Steinach mit über 3000 Einwohnern shop in.the village Steinach with over 3000 inhabitants verschwindet.¹⁶⁴ disappears ‘I cannot believe that it is true that the last big shop in the village Steinach is disappearing, even though over 3.000 people live there.’

(212)

Alles beginnt mit einem Blumenstrauß und einem everything begins with a flower.bouquet and a Wutanfall. Den Blumenstrauß hat Kerstin gepflückt – für ihre rage.attack the flower.bouquet has Kerstin collect-ppp for her Mama. Doch die sieht nur ungelenk abgerupfte Narzissen in der mum but she sees only awkwardly ripped daffodils in the Hand ihrer achtjährigen Tochter und fährt aus hand her-gen eight.year.old-gen daughter-sc gen and drives out der Haut: „Das darf ja wohl nicht wahr sein – meine the skin this is.allowed.to par maybe neg true be-inf my schönsten Gartenblumen, du spinnst wohl!”¹⁶⁵ beautiful-sup garden.flowers you nuts.be-inf maybe ‘Everything started with a bouquet of flowers and a fit of anger. The bouquet of flowers has been collected by Kerstin – for her Mum. But she only sees awkwardly ripped daffodils in the hand of her eight year old daughter and loses her temper: “I cannot believe that this is true – the most beautiful flowers from my garden, you are nuts!” ’

(213)

Der Norweger kommt trotz Aufwinds nur auf 111,5 the Norwegian comes in.spite.of updraught-gen only at 111,5 Meter hinunter. Für seinen Teamkollegen Velta darf das meters down for his team.colleague Velta is.allowed.to that kein Problem sein.¹⁶⁶ no problem be-inf ‘The Norwegian only jumped 111.5 meters, in spite of an updraught. I cannot imagine that this will be a challenge for his team mate Velta.’

However, the examples in (211) and (212) differ from well-behaved epistemic modal verbs in a crucial respect. In these cases, the speaker knows that the embedded proposition is actually true. This type of context is not compatible with epistemic operators. A similar observation regarding this use of dürfen has been already

164 DeReKo: A10/FEB.04341 St. Galler Tagblatt, 15/02/2010. 165 DeReKo: M11/MAI.06511 Mannheimer Morgen, 21/05/2011. 166 www.laola1.at 04/01/2012. Live Ticker for the 4-Schanzen-tournee.

2.2 Case studies |

121

made by Fritz (1991: 46 Fn.1). The peculiarity of these examples might be due to the fact that they do not carry an animate modal goal. The speaker does not want to believe the modified proposition, he does not permit himself to believe it. In more formal terms: The embedded proposition is not consistent with the wishes of the modal source, which is identified with the speaker. In a similar way, the interpretation of darf in (213) can be captured in terms of circumstantial modality. Being a well-trained ski jumper, it is not consistent with the wishes of the modal source, which is identified with the trainer. Similar abstract uses of circumstantial dürfen have been collected by Welke (1965: 107). As has been noticed by Öhlschläger (1989: 185), the semantic core of circumstantial dürfen and epistemic dürfte differ considerably. For the same reason, some authors such as Wurmbrand (2001: 137, 224) consider dürfte an independent lexical item. Further evidence for this view will be presented in Section 2.2.5. 2.2.4.3 Raising directionals with event modification As with the verbs that have been reviewed in the previous sections, deontic dürfen can be observed with verbless directional phrases. As has been pointed out by Barbiers (1995, 2002: 53) and Szumlakowski-Morodo (2006: 327), an ellipsis account particularly lacks plausibility in the cases in which the verb selects an inanimate NP. Such cases can be found with deontic dürfen. (214)

Wie viel Haar darf weg? Welche Farbe soll es how much hair is.allowed.to away what colour should it sein?¹⁶⁷ be-inf ‘How much hair am I allowed to cut off? What colour should it be?’

(215)

Außerdem heftete Jost einen Merkzettel dran, was in die Tonne moreover put Jost a reminder on.it what in the bin hinein darf und was nicht.¹⁶⁸ in is.allowed.to and what neg ‘Moreover, Jost put a reminder on it indicating what may be thrown in to the bin and what not.’

(216)

Einer darf noch in die WM-Abfahrt, a is.allowed still into the world.championship.downhill.race aber wer?¹⁶⁹ but who

167 DeReKo: M02/AUG.64829 Mannheimer Morgen, 31/08/2002. 168 DeReKo: M98/DEZ.93655 Mannheimer Morgen, 12/12/1998.

122 | 2 Modal verbs: A class struggle

‘Someone may still join the downhill race team for the world championship team, but who could it be?’

As the examples (214) and (215) do not contain subjects that can be identified as the bearer of a permission, they are most plausibly analysed as raising patterns. In contrast, example (216) involves a subject quantifier that takes narrow scope with respect to the modal operator, which is generally held to be a diagnostic for raising.

2.2.5 dürfte As has been illustrated in the previous section, the subjunctive of the past form dürfte can yield an epistemic interpretation. As it turns out, this use of dürfte occurs in the very same contexts as the remaining epistemic modal verbs occur. Just as with any other epistemic modal verb, the epistemic use of dürfte can embed a predication consisting of an identified individual and a predicate that refers to a state that cannot be changed (cf. 217–219) or a predicate that refers to an event in the past (cf. 220–221). A circumstantial interpretation in which dürfte is interpreted as the subjunctive of the past of deontic dürfen ‘be allowed to’ is not available in these cases. (217)

Das erste Bild des Babys dürfte Schätzungen von the first picture the-gen Baby-gen might estimations by Paparazzi zufolge rund fünf Millionen Dollar (vier papparazzi according.to about five millions dollar (four Millionen Euro) wert sein.¹⁷⁰ millions Euros) worth be-inf ‘According to estimations by paparazzi, the first picture might have a value of about five million dollars (four million euros).’

(218)

Fachleute haben inzwischen auch mit Scannern die experts have meanwhile also with scanners the Maße dieses neuen Hohlraumes gemessen. Er dürfte measurements the-gen new-gen cavity-gen measured he might 15 mal 13 Meter groß und 48 Meter tief sein.¹⁷¹ 15 times 13 meters big and 48 meters deep be-inf ‘Meanwhile, experts have also measured the dimensions of the new cavern with scanners. It is 15 by 13 meters wide and 48 meters deep.’

169 DeReKo: K97/FEB.08651 Kleine Zeitung, 04/02/1997. 170 DeReKo: NUZ06/MAI.02995 Nürnberger Zeitung, 29/05/2006. 171 DeReKo: NON07/JUN.12389 Niederösterreichische Nachrichten, 20/06/2007.

2.2 Case studies |

(219)

123

Überhaupt hatte der Pudel, er dürfte schätzungsweise neun bis generally had the poodle he might approximately nine to zehn Jahre alt sein, Glück im Unglück.¹⁷² ten years old be-inf luck in.the bad.luck ‘Generally, the poodle, who is approximately nine or ten years old, was quite lucky, given the circumstances.’

(220)

Der Mann dürfte im Schlaf gestorben sein, da die Beamten the man might in.the sleep die-ppp be-inf as the officers ihn im Bett gefunden hatten.¹⁷³ him in.the bed found had ‘The man must have died in his sleep, as the officers had found him in his bed.’

(221)

Der Wunsch nach Ungestörtheit dürfte schließlich dem Liebespaar the wish for privacy might finally the love.couple auf so tragische Weise das Leben gekostet haben.¹⁷⁴ for such tragic manner the live cost-ppp have-inf ‘The wish for privacy might have finally caused the death of the lovers, who died in such a tragic manner.’

Furthermore, this use of dürfte is subject to the CoDeC. In all of the examples above, the speaker indicates that the modified proposition is not part of his knowledge. For instance, the author of example (217) could not resume the discourse with an utterance such as ‘. . . and indeed, a newspaper has paid 4.8 million dollars for that picture.’ It is no trivial matter to identify the precise semantic specification of this use. Some authors conclude that dürfte equals the epistemic uses of können: Bech (1949: 20, 38) assumes that epistemic dürfte has the same modal force as epistemic können, and Lötscher (1991: 353) argues that epistemic dürfte has the same meaning as epistemic könnte. In contrast, Welke (1965: 107) observes that dürfte cannot be analysed compositionally. According to him, it behaves similarly to the epistemic use of können, but it expresses a higher degree of certainty. This perspective has been adopted by Raynaud (1977: 23), Weinrich (1993: 312), Zifonun (1997: 1910), Öhlschläger (1989: 195, 258) and Helbig and Buscha (2001: 121), who argue that epistemic dürfte expresses a probability. Likewise, Kratzer (1981: 58) notices that epistemic dürfte is hard to gloss. According to her, the most promising translation is it is

172 DeReKo:RHZ06/OKT.11580 Rhein-Zeitung, 13/10/2006; Pudel einfach über den Zaun geworfen 173 DeReKo: BVZ09/OKT.01155 Burgenländische Volkszeitung, 14/10/2009. 174 DeReKo: NON09/JAN.04467 Niederösterreichische Nachrichten, 12/01/2009.

124 | 2 Modal verbs: A class struggle

probable that. In subsequent work, Kratzer (1991: 650) classifies dürfte as a weak necessity modal verb. In a similar manner, Becker (1836: 180) and Schoetensack (1856: 297) observed already that epistemic dürfte expresses a logical probability. Both of them assume that this interpretation has been derived from morphologically similar preterite present turren ‘dare’. Furthermore, they conclude that the speaker’s evaluation is dependent on the volition of another agent. Similarly, Curme (1922: 319) argues that dürfte is used to state that the speaker is pretty sure about the validity of the embedded proposition. Finally, Vater (1975: 112) notices that epistemic dürfte always involves a weaker type of modal force than epistemic werden. As none of these authors provides sound empirical evidence for their classifications, the semantic behaviour of dürfte will receive closer attention in the remainder of this section. As was shown in the preceding section, there are a couple of diagnostics that apply to well-behaved possibility modal verbs. Firstly, departing from the hypothesis advocated by Levinson (2000: 36), according to which epistemic possibility operators induce scalar implicatures, Papafragou (2006: 1693) and Kotin (2008: 382) argue that a canonical epistemic possibility operator should not cause a contradiction in a configuration in which the possibility operator is conjoined with the possibility operator that selects the negated proposition such as: (♢p) & (♢¬p). And indeed, the epistemic possibility verbs kann and könnte are acceptable in this type of configuration, as is illustrated in examples (222a)–(222b). However, epistemic dürfte yields a contradiction in such contexts (cf. 222c): (222)

a.

Anatol kann den Brief gelesen haben, aber er kann ihn Anatol can the letter read-ppp have-inf but he can it genauso gut auch || nicht gelesen haben. exactly.as well also neg read-ppp have-inf ‘It could be that Anatol has read the letter, but it could also be that he has not read it.’

b.

Anatol könnte den Brief gelesen haben, aber er Anatol can-sbjv.pst the letter read-ppp have-inf but he könnte ihn genauso gut auch || nicht gelesen haben. can-sbjv.pst it exactly.as well also neg read-ppp have-inf ‘It could be that Anatol has read the letter but it could also be that he has not read it.’

c.

# Anatol dürfte

den Brief gelesen haben, aber er Anatol may-sbjv.pst the letter read-ppp have-inf but he dürfte ihn genauso gut auch || nicht gelesen haben. can-sbjv.pst it exactly.as well also neg read-ppp have-inf Intended reading:‘It is more than likely that Anatol has read the letter but it is also more than likely that he has not read it.’

2.2 Case studies |

125

As it seems, the most natural prosodic pattern for these configurations is one in which the focus is on the past participle gelesen in the first clause, and on the negation in the second clause. The resulting set of alternatives is reminiscent of VERUM-focus. What is under debate is whether Anatol has read the letter or not. Crucially, epistemic dürfte behaves significantly differently in these configurations from its deontic counterpart, which does not cause a contradiction. Upon closer inspection, it turns out that dürfte does not trigger scalar implicatures at all under the same prosodic pattern (cf. 222c) . This is in opposition to the canonical epistemic possibility modal verbs kann (cf. 222a) and könnte (cf. 222b), which behave exactly as Levinson (2000: 36) would expect. (223)

a.

Anatol kann den Brief gelesen haben, aber er muss ihn Anatol can the letter read-ppp have-inf but he must it nicht gelesen haben. neg read-ppp have-inf ‘It could be that Anatol has read the letter but it is not certain that he has read it.’

b.

Anatol könnte den Brief gelesen haben, aber er muss Anatol can-sbjv.pst the letter read-ppp have-inf but he must ihn nicht gelesen haben. it neg read-ppp have-inf ‘It could be that Anatol has read the letter but its not certain that he has read it..’

c.

# Anatol dürfte

den Brief gelesen haben, aber er muss Anatol may-sbjv.pst the letter read-ppp have-inf but he must ihn nicht gelesen haben. it neg read-ppp have-inf

Intended reading:‘It is more than probable that Anatol has read the letter but it need not be that he has read it.’

It appears that the acceptability of example (222c) would increase if the accent were on dürfte, but this does not change the fact that there is a clear contrast between the examples with könnte and those with dürfte. The reason for the incompatibility of epistemic dürfte remains mysterious. If it is indeed a verb that expresses a probability, the behaviour in example (222c) is unexpected. In this case, the first clause would express that the likelihood that p is true is higher than 0.5, and the second clause would express that the likelihood for p is lower than 1.0. From a merely logical perspective, this does not yield any contradiction. The contrasts become more obvious as soon as the order of the conjuncts is changed and the focus targets the modal force of the modal operator.

126 | 2 Modal verbs: A class struggle

(224)

a.

Anatol muss den Brief nicht gelesen haben, aber er kann Anatol must the letter neg read-ppp have-inf but he can ihn gelesen haben. it read-ppp have-inf ‘Anatol has not necessarily read the letter but it could be that he has read it.’

b.

Anatol muss den Brief nicht gelesen haben, aber er Anatol must the letter neg read-ppp have-inf but he ihn gelesen haben. könnte can-sbjv.pst it read-ppp have-inf ‘Anatol has not necessarily read the letter but it could be that he has read it.’

c.

# Anatol muss den Brief nicht gelesen haben,

aber er Anatol must the letter neg read-ppp have-inf but he ihn gelesen haben. dürfte may-sbjv.pst it read-ppp have-inf

Intended reading: ‘Anatol has not necessarily read the letter but it is more than probable that he has read it.’

All of the examples considered so far have illustrated, that dürfte cannot be considered a well-behaved epistemic possibility modal verb. However, if dürfte were a probability modal verb, as is often assumed, the contrasts in example (223) and (224) would be mysterious. As it seems, by using dürfte the speaker makes a commitment that he considers the embedded proposition among a set of alternatives. But this cannot be the whole story. Obviously, it involves some additional semantic features which prevent the modal force of dürfte from being contrasted with epistemic muss. This feature may involve some evidential dimension, as is sometimes claimed, but the exact nature of this feature remains to be investigated. Following Huitink’s (2008) findings, there are epistemic uses that can occur in the scope of a quantifying NP. As will be shown in Section 4.20, the most productive epistemic modal verbs in this configuration are the possibility modal verbs können and könnte. The universally quantifying NP can clearly bear scope over können (cf. 225a) and könnte (cf. 225b), yielding an interpretation in which the culprit could be identified with any person. Such a reading is not available with dürfte (cf. 225c): In this case, the narrow scope interpretation prevails, referring to an implausible state of affairs in which the culprit is everybody at the same time. (225)

a.

Dieses Schaufenster kann jeder eingeschlagen haben this shop.window can everybody break-ppp have-inf ‘Anybody could have broken this shop window.’

2.2 Case studies |

b.

127

Dieses Schaufenster könnte jeder eingeschlagen this shop.window can-sbjv.pst everybody break-ppp haben have-inf ‘Anybody could have broken this shop window.’

c.

# Dieses Schaufenster dürfte

jeder eingeschlagen this shop.window may-sbjv.pst everybody break-ppp haben have-inf Intended reading: ‘Probably everybody has broken this shop window.’

The same contrasts arise with the examples inspired by Huitink (2008), and the clauses based on the corpus examples (317) and (319): (226)

Mindestens drei Männer können der Vater meines Kindes at.least three men can the father my-gen child-gen sein. be-inf ‘At least three different men could be the father of my child.’

(227)

# Mindestens drei Männer dürften der Vater meines Kindes

at.least sein. be-inf

three men

may

the father my-gen child-gen

Intended reading: ‘Perhaps, at least three different men are the father of my child.’

(228)

Da die Schule den ganzen Tag und teilweise auch am späteren as the school the whole day and partially also at later Abend zugänglich ist, kann jeder die Kopien mitgenommen evening accessible is can everyone the copies with.take-ppp haben.¹⁷⁵ have-inf ‘As the school is open all day and sometimes until late in the evening, anyone could have taken the copies.’

(229)

# Da die Schule den ganzen Tag und teilweise auch am späteren

as the school the whole day and partially also at later Abend zugänglich ist, dürfte jeder die Kopien mitgenommen evening accessible is might everyone the copies with.take-ppp haben. have-inf

175 DeReKo: A98/JUN.37190 St. Galler Tagblatt, 05/06/1998.

128 | 2 Modal verbs: A class struggle

Intended reading: ‘As the school is open all day and sometimes until late in the evening as well, perhaps everyone has taken the copies.’

(230)

„Diesen Brief könnte jeder geschrieben haben, es geht in this letter could everyone write-ppp have-inf it goes in keine politische Richtung”, so Werner Hosiner-Gradwohl.¹⁷⁶ no political direction so Werner Hosiner-Gradwohl ‘ “Anyone could have written this letter; it does not indicate any political direction”, according to Werner Hosiner-Gradwohl.’

(231)

# „Diesen Brief dürfte jeder

geschrieben haben, es geht in this letter might everyone write-ppp have-inf it goes in keine politische Richtung”, so Werner Hosiner-Gradwohl. no political direction so Werner Hosiner-Gradwohl

‘Intended reading: “Everyone could have written this letter: It does not indicate any political direction”, according to Werner Hosiner-Gradwohl.’

To conclude, epistemic dürfte does not behave like the prototypical epistemic possibility verbs können and müssen in a number of respects. Thus, it cannot be considered an epistemic possibility verb. It is obvious that it carries a modal force that is stronger than that. Therefore, the widespread analysis of epistemic dürfte as a probability modal verb is more plausible. But, as has been shown, this type of analysis does not capture the semantic behaviour in all respects. The epistemic use of dürfte additionally involves a semantic component that yet remains to be identified. Vater (1975: 112) claims that it always contributes an ironical resonance. Whether this is the key to the mysteries described above remains to be seen. Considering the different uses of modal verbs, it turns out that alternations of the modal force between different uses of a particular verb hardly ever occur. The only prominent case concerns the verb mögen and its subjunctive of the past form möchte: Whereas the former is generally held to carry a modal possibility operator, the latter is by and large synonymous with the volitional verb wollen and therefore most plausibly analysed as necessity modal verb. Crucially, möchte is canonically considered an independent lexical item that has emancipated from its host lexicon entry mögen, as is illustrated in some detail by Öhlschläger (1989: 7), Kiss (1995: 162), Fritz (1997: 103), Diewald (1999: 144), Wurmbrand (2001: 137, 224) and Vater (2010: 103). A more detailed discussion is given in Section 2.2.8. In a similar fashion, it is reasonable to follow Wurmbrand (2001: 137, 224), who argues that dürfte is an independent lexical item. If there is a common semantic core for deontic dürfen and epistemic dürfte, it must be very minor. If this is true, any account, such as Diewald (1999: 1) or Reis (2001: 287), that defines

176 DeReKo: BVZ07/SEP.03009 Burgenländische Volkszeitung, 26/09/2007.

2.2 Case studies |

129

the class of modal verbs in German in terms of poly-functionality faces a great challenge. If deontic dürfen and epistemic dürfte cannot be subsumed under one lexical entry, there are suddenly two mono-functional items in the class of modal verbs. Finally, it merits closer attention that epistemic dürfte occurs in a large range of environments in which most other epistemic modal verbs cannot be found, such as information-seeking question and embedded clauses, as will be seen in in Chapter 4.

2.2.6 sollen Beginning with Becker (1836: 181), sollen is considered as a modal verb that indicates external volition. In crucial respects, it very much resembles the volitional verb wollen. This idea has been further proliferated by Bech (1949: 11). On the one hand, its (quasi) deontic use always requires an animate source of modality. In the case of sollen, this volitional modal source is not realised as the subject NP, but it remains syntactically unrealised and implicit. On the other hand, sollen obtains a reportative interpretation just under the same conditions as wollen does. Moreover, it exhibits an idiosyncratic behaviour with respect to negation. Apart from this, sollen can yield a truly epistemic reading, as soon as it bears the subjunctive of the past morphology. Finally, sollen could be used as a transitive verb until the Early New High German period. 2.2.6.1 Transitive Uses In earlier stages of German, sollen was used as a transitive verb that denoted a debt. The examples DWB provides the following examples from Old High German and Middle High German: (232)

Zuene sculdigon uuarun sihuuelihemo inlihere: ein solta two debtor were anybody-dat lender-dat one shall-pst finfhunt pfenningo, ander solta finfzug.¹⁷⁷ five.hundred pennies other shall-pst fifty ‘A money lender had two debtors: one of them owed him five hundred pennies, the other one fifty.’

(233)

swer im iht sol, der mac wol sorgen¹⁷⁸ whoever him something shall the may well care-inf

177 Tatian 138, 9 (830).

130 | 2 Modal verbs: A class struggle

‘Whoever owes him something might care about that.’

Contributing a highly specific semantics, it is reasonable to assume that the transitive use of sollen represents the source from which the subsequent necessity modal verb developed. A debt is nothing but a very specific necessity. By means of semantic bleaching, a predicate that expresses a debt can easily turn into a predicate that expresses an obligation or a more abstract necessity. According to the DWB (p. 1469), the transitive use of sollen has only disappeared in the course of the New High German period. Denison (1993: 306) provides analogous examples for shal in Middle English. Apart from that, a similar situation can be found in French. In contemporary language use, the most canonical necessity verb devoir ‘must, shall’ is still occasionally used as a transitive verb that refers to a debt, as is illustrated in (234): (234)

Je dois 51 euros à mon épicier. I shall 51 euros to my greengrocer ‘I owe 51 euros to my greengrocer.’

Even if the transitive use of sollen has disappeared in Contemporary German, there are instances that could be considered as transitive-like patterns, but their usage is strictly restricted. The case under discussion here is restricted to wh-questions and idiomatic. Interestingly, this pattern is less acceptable with other traditional modal verbs that cannot be used as transitive verbs today, such as dürfen and müssen. (235)

Was soll/? darf/*muss ich denn hier?¹⁷⁹ what shall/am.allowed.to/must I par here ‘What am I supposed to do here?’

As these uses are fairly restricted and highly idiomatic, they will not receive any further attention. 2.2.6.2 Raising infinitives with event modification The most frequent use of sollen takes a bare infinitive complement and denotes a volition that is attributed to a referent different from the subject referent, as 178 Walther von der Vogelweide Frô Welt, ir sult dem wirte sagen 100, 28 (around 1200). 179 The example with dürfen gets slightly more acceptable if the particles überhaupt noch ‘at.all’ are added. However, müssen remains ungrammatical under the same conditions: (1)

Was darf/*muss ich denn hier überhaupt noch? what am.allowed.to/must I par here at.all still ‘What am I allowed to do here at all?’

2.2 Case studies |

131

has already been indicated by Bech (1949: 11), Raynaud (1977: 14), Glas (1984: 45), Öhlschläger (1989: 174) and Fritz (1997: 17). In terms of meaning, sollen is very close to the obligation reading of müssen. It is thus reasonable to consider both elements as deontic modal verbs. The major difference concerns the specification of the modal source, which is less restrictive in the case of müssen and always confined to an animate agent, who is usually identified in the prior discourse. Authors such as Höhle (1978: 87) suggest that sollen even contributes an implicit argument for this bearer of volition. Other more idiosyncratic uses of circumstantial sollen are discussed by Welke (1965: 98), Bech (1949: 13–18) and Glas (1984). Hinterwimmer (2014) points out that the semantics of sollen is even more specific: it does not only presuppose a volitional agent but the proposition which contains sollen asserts the existence of a intentional act. This intentional act involves a bearer of the volition which is distinct from the subject referent and a speech act in which that bearer of expresses a wish. This is reflected by the fact that sollen behaves differently from imperatives in a crucial aspect. Whereas imperatives can be used to directly refer to an addressee (236), sollen with a 2nd person subject cannot (237). (236)

Geh jetzt! go-imp now ‘Go now!’

(237)

# Du sollst jetzt gehen!

you shall now go-inf ‘You shall go now!/I want you to go now!’

(237) can only felicitously be uttered if it is settled in the common ground between the speaker and the addressee that the speaker wants the addressee to leave, such as a previous utterance of an imperative like (236). This restriction is reminiscent of the English semi-modal have with to-infinitive. As observed by Perkins (1983: 60) the modal source for have to cannot be the speaker. It seems that Hinterwimmer’s observation is rooted in the semantics of wollen (cf. Section (2.2.3). Moreover it is reflected by the semantics of reportative sollen and, to some extent, by the semantics of epistemic sollte. The requirement of an intentional act prior to utterance time could also be due to the original meaning of sollen which used to refer to a state of being indebted. With respect to its argument structure, the precise nature of circumstantial sollen is contested just as most circumstantial modal verbs are. On the one hand, there are authors such as Welke (1965: 87) and Abraham (2002: 38), who explicitly assume that circumstantial sollen carries a referential subject argument of its own that identifies the person on which the obligation is imposed. On the other hand,

132 | 2 Modal verbs: A class struggle

other authors, such as Öhlschläger (1989: 105) and Wurmbrand (2001: 187–204), provide evidence that sollen involves a raising pattern. Applying the diagnostics presented in the previous sections, it turns out that circumstantial sollen is indeed a raising verb: It occurs with non-referential subjects (cf. 238–239) and it permits de dicto interpretations of quantifiers in subject position, in which the modal operator bears scope over the quantifier (cf. 240–241). At this point, it is not important whether ein is used as an indefinite pronoun or as a numerical determiner, as Carpenter (1998: 87) has illustrated that numerical determiners behave in the same manner as ordinary existential quantifiers. (238)

Die Zielsetzung ist für beide Teams klar: Es soll gepunktet the goal is for both teams clear: it shall score-ppp werden.¹⁸⁰ pass.aux-inf ‘The goal is evident for both teams: They are supposed to score.’

(239)

Auch in Salzgitter forderten gestern viele Menschen, dass es also in Salzgitter demand yesterday many people that it nie wieder Krieg geben soll.¹⁸¹ never again war give-inf shall ‘In Salzgitter, many people also demanded that there should not be any war any more.’

(240)

Ein Kandidat „von außen” soll das Gerangel um die a candidate from outside shall the bully around the Chefredakteursfunktion im ORF-Landesstudio chef.editor.function in.the ORF-regional.television.studio beenden.¹⁸² terminate-inf ‘A candidate from outside shall terminate the bully around the position of the chief editor in the regional television of the ORF.’

(241)

„Ein Beamter von dort soll die Reichenau verstärken”, an officer from there shall the Reichenau reinforce-inf fordert Strigl.¹⁸³ demanded Strigl ‘ “An officer from down there shall reinforce the department in Reichenau”, demanded Strigl’

180 DeReKo: BVZ11/MAI.00930 Burgenländische Volkszeitung, 12/05/2011. 181 DeReKo: BRZ08/SEP.01065 Braunschweiger Zeitung, 02/09/2008. 182 DeReKo: K99/SEP.67351 Kleine Zeitung, 09/09/1999. 183 DeReKo: I97/JUL.27921 Tiroler Tageszeitung, 19/07/1997.

2.2 Case studies |

133

In a similar way, circumstantial sollen turns out to be transparent with respect to voice. If sollen were a control verb, the obligation should be expected to always be imposed on the subject argument. Accordingly, the bearer of the obligation should be Reinhold in (242a) and the Nanga Parbat in (242b). As the latter is a mountain and therefore no licit bearer of obligation, this interpretation is not plausible for the examples in (242). Instead, both examples refer to the same state of affairs, they only differ with respect to their information structure. The bearer of obligation is not identified by the assignment of a semantic role. (242)

a.

Der Reinhold soll den Nanga Parbat ohne the Reinhold shall the-acc Nanga Parbat without Sauerstoffgerät bezwingen. oxygen.apparatus conquer ‘It is allowed that Reinhold conquers the Nanga Parbat without oxygen mask.’

b.

Der Nanga Parbat soll vom Reinhold ohne the-nom Nanga Parbat shall by.the Reinhold without Sauerstoffgerät bezwungen werden. oxygen.apparatus conquer-prt.pas pas.aux.pst It is allowed that the Nanga Parbat is conquered by Reinhold without oxygen mask.”

As has been noticed by Bech (1951: 8), Öhlschläger (1989: 91), Fritz (1997: 55), Zifonun (1997: 1903) and Ehrich (2001: 162), circumstantial sollen exhibits a peculiar behaviour with respect to the scope of negation. These authors argue that circumstantial sollen seems to prefer a narrow scope interpretation whilst all of the remaining traditional circumstantial modal verbs prefer a wide scope interpretation. Assuming that sollen expresses an obligation, the canonical interpretation of the negation in cases such as (243) is as follows: ‘it is mandatory not to commit adultery’, rather than ‘it is not mandatory to commit adultery’. As already pointed out in Section 2.2.4.2, the diacritic || indicates a intonation break and the underlined constituent bears a high pitch accent. (243)

Du sollst nicht ehebrechen. you shall neg commit.adultery-inf ‘Thou shalt not commit adultery.’

(244)

Lola soll ihre Mutter nicht ärgern. Lola shall her mother neg annoy-inf ‘Lola shall not annoy her mother.’

(245)

Sie darf ihre Mutter auch || nicht ärgern. Lola is.allowed.to her mother also neg annoy-inf ‘Alternatively, Lola is also allowed not to annoy her mother.’

134 | 2 Modal verbs: A class struggle

(246)

Sie soll ihre Mutter || nicht ärgern. Lola shall her mother neg annoy-inf ‘What Lola shall do is not to annoy her mother.’

(247)

Nicht anrufen soll man auch nicht¹⁸⁴ neg call shall one also neg ‘What you should not do either, is not to call.’

However, upon closer inspection, the situation turns out to be much more complicated. Even if it is more plausible to interpret the negation in (243) and (244) with a narrow scope relative to sollen, there are a whole range of aspects that would remain unaccounted for with such a perspective. Firstly, a narrow scope interpretation of a negation that is combined with a circumstantial modal verb only becomes available in German under a marked prosodic pattern, as has been pointed out by Blühdorn (2012: Sect. 8.5). The negative particle requires a high pitch accent and needs to be set off by an intonational break, as is demonstrated in example (245). This prosodic pattern induces a contrast focus on the negation. The resulting set of alternatives is {‘it is allowed that Lola annoys her mother’, ‘it is allowed that Lola does not annoy her mother’}. Contrasting the negation, the configuration in example (245) presupposes that one of the discourse participants suggested to add the positive proposition ‘it is allowed that Lola annoys her mother’ to the common ground. Some sort of VERUM focus is at work. This is the only way how a negation can occur in the scope of a circumstantial modal verb in German. If the speaker has the choice to utter the logically equivalent alternatives ‘Lola darf ihre Mutter nicht ärgern’ and ‘Lola muss ihre Mutter nicht ärgern’, he would, in most cases, choose the latter one, as it does not impose as many contextual restrictions as the former one. Turning to the most typical cases of sollen, occurring with negation, such as in examples (243)–(244), it becomes clear that they behave differently. On the one hand, these patterns involve an unmarked intonation in which the negative particle does not receive a contrastive focus stress. On the other hand, these sentences can be uttered even if the positive proposition such as ‘Lola soll ihre Mutter ärgern’ has not been added to the prior discourse. In order to obtain this effect, the negative particle requires being stressed, as in example (246). The canonical case of sollen with negation exhibits striking similarities to the remaining circumstantial modal verbs which occur in the scope of the negation in the unmarked case. This is on par with the observation made by Öhlschläger (1989: 91), who noticed that sollen can occur with two negative particles (cf. 247).

184 As quoted in Öhlschläger (1989: 91).

2.2 Case studies |

135

From this it follows that sollen should be possible with a wide scope interpretation as well. Moreover, the negative particle can remain in the same clause as sollen, in a wh-cleft configuration in which the infinitive complement occurs in the other clause, as is illustrated in (248)–(249). As Lenz (1996: 416) argues, wh-clefts can be used to disambiguate the scope of the negation with respect to modal verbs, which carry different interpretations in written German otherwise. More details on this pattern are discussed in Section 4.8. (248)

Was wir aber nicht sollten, ist Bürgern vorwerfen, daß sie what we but neg shall-sbjv.pst is citizen blame-inf that they ihre Vergangenheit nicht bewältigt hätten.¹⁸⁵ their past neg overcome-ppp have-sbjv.pst ‘What we should not do is to blame citizens for not having come to terms with their past.’

(249)

Was Kunst aber nicht sollte, ist sich aus der Frage, what art but neg shall-sbjv.pst is refl out the question welche ästhetischen Mittel angemessen sind, einfach what aesthetic means appropriate are simply herauszulügen.¹⁸⁶ out.to.lie-inf ‘What art should not do is to avoid the issue which aesthetic means are appropriate.’

How can the contradictory facts concerning the interpretation of sollen with respect to negation be reconciled? As for the interaction of the necessity operator and the negation, a narrow scope interpretation seems to be much more plausible. However, from a syntactic and prosodic perspective, a wide scope interpretation is the more appropriate one. Assuming that sollen carries an external volition, the riddle can be solved without much further ado. Note that for the volitional verbs such as wollen, the difference between a wide scope interpretation of a negation and a narrow scope interpretation can be very minor: ‘I do not want that you feel bad’ versus ‘I want that you do not feel bad’. Moreover, a wide scope reading is often re-interpreted as narrow scope reading by means of an implicature (cf. ‘NEG raising’). Likewise, it is plausible to assume that a negation that takes wide scope over sollen is not interpreted as a negation applied to a necessity ¬□p or ‘it is not necessary that p’. Rather, it interacts with the volitional modal operator. A more appropriate translation is: ‘there is some referent to which it is attributed that

185 DeReKo: RHZ97/FEB.14043 Rhein-Zeitung, 24/02/1997. 186 DeReKo: HAZ08/NOV.04835 Hannoversche Allgemeine, 27/11/2008.

136 | 2 Modal verbs: A class struggle

he does not want p to happen. In contrast, deontic müssen in the scope of negation could be translated as: ‘there is no referent to which it is attributed that he wants p to happen’. As will be demonstrated in Section 4.10, there are different options of how a negative operator can interact with modal operators. It need not always affect the entire modal operator, occasionally only some components are concerned. 2.2.6.3 Raising directionals with event modification Just like its circumstantial relatives, sollen is frequently found with verbless directional phrases. As has been illustrated in Section 2.2.1, patterns with inanimate subjects provide strong evidence for an underlying raising pattern. Interestingly, sollen can be found more frequently in configurations without animate subjects, as compared to the necessity modal verb müssen. (250)

Die „Blechdose” am Alsteranleger Alte Rabenstraße soll the Blechdose at.the Alster.quay Alte Rabenstraße shall weg.¹⁸⁷ away ‘The “Blechdose” at the Alster quay Alte Rabenstraße shall disappear.’

(251)

Nicht selten fällt dann aber der Entscheid: das Rad muss weg, ein neg rarely falls then but the decision the bike must away a neues, moderneres soll her.¹⁸⁸ new modern-comp shall here ‘In much cases, the decision is: The bike has to be scrapped, a new one has to be acquired.’

(252)

Der Dreck der vergangenen Monate soll weg, die the dirt the-gen past-gen month-gen shall away the Frühlingssonne durch klare, streifenfreie Scheiben scheinen.¹⁸⁹ spring.sun through clear stain.less windows shine-inf ‘The dirt of the past months has to disappear and the spring sun should shine through clear and stainless windows.’

(253)

dieses Stück Demokratie soll nicht weg¹⁹⁰ this piece democracy shall neg away ‘This piece of democracy should not disappear.’

187 188 189 190

DeReKo: HMP05/APR.00259 Hamburger Morgenpost, 16/04/2005. DeReKo: A01/MAR.12763 St. Galler Tagblatt, 30/03/2001. DeReKo: BRZ07/MAR.20931 Braunschweiger Zeitung, 13/03/2007. DeReKo: Hannoversche Allgemeine, 01/10/2007.

2.2 Case studies |

137

Some of these examples reveal interesting patterns. The instance of sollen in (253) occurs next to a negation. A theory that assumes that the negation always takes narrow scope with respect to sollen would have to account for the precise nature of the element over which it bears scope in the example given above. In example (252), the verbless directional phrase occurs in coordination with an infinitive complement. Following a common assumption, the two conjuncts of a coordination have to be of the same category or ‘rank’, cf. Dougherty (1970: 850, 864), Jackendoff (1977: 51), Gazdar (1981: 157, 173), Schachter (1984: 269) or Pollard and Sag (1994: 202) for discussion. This could be interpreted as a hint that the verbless directional phrase contains a phonologically empty infinitive. But alternatively, one could argue that what is conjoined in (252) are two predicates. In earlier stages of German, sollen could even select a non-referential subject in configurations with verbless directional phrases, as examples (254)–(255), taken from the DWB, illustrate. (254)

angesehen, wie gar ein geringes leiden es ist, wenn es considered how intn a small suffering it is if expl gleich zum tode oder sterben sollt.¹⁹¹ directly to death or die-inf shall ‘Considering the fact of how little you suffer if you are about to die.’

(255)

nun soll es gerade auf Inspruck¹⁹² now shall expl straigth to Innsbruck ‘Now, they want us to go straight to Innsbruck’

But it is important to stress that these instances are not acceptable in contemporary Standard German. 2.2.6.4 Raising infinitives with clause modification: reportative Just like wollen, sollen can refer to a claim. This has been already observed by Becker (1836: 181), Schoetensack (1856: 295), Curme (1922: 322) and Bech (1949: 13). As pointed out above, circumstantial sollen can be semantically derived from circumstantial/volitional wollen, as it expresses volition that is attributed to a referent different from the subject referent. In the same manner, reportative sollen can be derived from reportative wollen referring to a claim that is attributed to a referent different from the subject referent. This has already explicitly been suggested by Schoetensack (1856: 295), who argues that reportative sollen involves a kind

191 Luther, br 4, 257 (ca. 1530) as it is quoted by the DWB. 192 Goethe XXVII, 11; (ca. 1800), as it is quoted by the DWB.

138 | 2 Modal verbs: A class struggle

of ellipsis as well: “dieser Mensch soll gesagt haben – jemand, irgendein Fremder verlangt, dass man glaube, dieser Mensch habe gesagt” (‘Somebody wants (that one believes) that this person has said . . . ’). Regarding the argument structure of sollen, the analysis suggested by Welke (1965: 97) is without opposition; reportative sollen is generally held to be a one-place predicate. Likewise, Bech (1949: 13) has already observed that canonical volitional sollen targets the “realisation” of the embedded predication. The uses above however target the “reality” of the embedded predication. This contrast corresponds exactly to the one between circumstantial interpretations of können or müssen and their epistemic counterparts. Moreover Hinterwimmer (2014) showed that sollen asserts the existence of an intentional act in the commonground. In the case of reportative sollen this intentional act is interpreted as a claim. Once again, these observations are in line with Truckenbrodt’s (2006: 263– 268) assumption, according to which any sentential speech act type carries a volitional element which is associated with the speaker. In uttering an assertion, the speaker expresses that he wants the addressee to add the proposition to the Common Ground. Being a volitional verb, reporative sollen expresses this volitional component of declarative clauses and associates it with a contextually given referent referent, which is different from the speaker. Similarly to wollen, sollen exhibits a behaviour parallel to epistemic modal verbs in a crucial respect. Whenever sollen embeds a predication consisting of an identified individual and a predicate that refers to a state that cannot be changed (cf. 256–257), or a predicate that refers to an event in the past (cf. 258–263), a circumstantial interpretation is excluded. The only reading available is the reportative one: (256)

Tom Cruise und Katie Holmes sind geschockt. L. R. Hubbard (kl. Tom Cruise and Katie Holmes are shocked L R Hubbard small F.) soll Suris Vater sein.¹⁹³ picture shall Suri-gen father be-inf ‘Tom Cruise and Katie Holmes are shocked. L. R. Hubbard is claimed to be Suri’s father.’

(257)

So soll das Kokain einen Reinheitsgehalt von 80 bis 90 Prozent so shall the cocaine a purity.degree of 80 to 90 percent besitzen und ca. 40 Millionen Euro wert sein.¹⁹⁴ have and about 40 million Euro worth be-inf

193 DeReKo: HMP08/JAN.00616 Hamburger Morgenpost, 08/01/2008.

2.2 Case studies |

139

‘According to this, the cocaine is claimed to have a purity degree of around 80 or 90 percent and worth about 40 million Euro.’

(258)

Horst Seehofers Ex-Geliebte Anette Fröhlich (35) soll den Horst Seehofer-gen ex-lover Anette Fröhlich 35 shall the CSU-Chef laut „Bunte” zuletzt in Berlin „regelmäßig und CSU-head according.to Bunte recently in Berlin regularly and lange” besucht haben.¹⁹⁵ long visit-ppp have-inf ‘According to the “Bunte”, Horst Seehofer’s ex-lover Anette Fröhlich is claimed to have recently visited the CSU head in Berlin frequently and for long periods.’

(259)

Während eines Gottesdienstes soll sie sich einmal an seinem during a church.service shall she refl once on his Messgewand festgekrallt haben. Im Beichtstuhl soll liturgical.vestment clinge-ppp have-inf in.the confessional shall es sogar zu einem „Annäherungsversuch” gekommen sein.¹⁹⁶ it even to a advance come-ppp be-inf ‘During a church service she is claimed to have clinged to his liturgical vestment. She is said to have made an advance in the confessional.’

(260)

Im Sommer 2008 soll Kaczynski auf dem Flug in das von in summer 2008 shall Kaczynski on the flight to the by Russland bedrängte Georgien mit einem Wutanfall auf die Russia harried Georgia with a rage.attack on the Entscheidung des Piloten reagiert haben, aus decission the-gen pilot react-ppp have-inf for Sicherheitsgründen im benachbarten Aserbaidschan zu landen. security.reasons in neighbouring Azerbaijan to land Später hatte er dem Piloten Feigheit vorgeworfen und soll after had he the pilot cowardice blame-ppp and shall seine Entlassung erwirkt haben.¹⁹⁷ his dismissal obtain-ppp have-inf ‘Reportedly, Kaczynski reacted during the flight to Georgia with a rage attack, having acknowledged the decision of the pilot to land in the neighbouring country Azerbaijan for security reason. It is further claimed that he accused the pilot of being a coward and has obtained his dismissal.’

194 195 196 197

DeReKo: NON10/FEB.11326 Niederösterreichische Nachrichten, 17/02/2010. DeReKo: HMP09/JUN.01135 Hamburger Morgenpost, 14/06/2009. DeReKo: A09/FEB.04884 St. Galler Tagblatt, 18/02/2009. www.orf.at, accessed on 27th May 2010.

140 | 2 Modal verbs: A class struggle

(261)

Die Schweizer Journalistin Klara Obermüller, bei der Guttenberg The Swiss journalist Klara Obermüller, from the Guttenberg in seiner Doktorarbeit abgeschrieben haben soll, findet dieses in this thesis copy-ppp have-inf shall finds this Verhalten „nicht sehr ehrenhaft und eigentlich auch nicht sehr behavior neg very honourable and actually also neg very klug”.¹⁹⁸ smart ‘The Swiss journalist from whom Guttenberg copied some passages in his thesis considers this behaviour “not very honourable and actually also not very smart”.’

(262)

Nordwestlich von Pjöngjang sollen drei Wildgänse northwest of Pyongyang shall three wild.gooses beobachtet worden sein, die ein Trauergeschrei von observed-ppp pass.aux-ppp be-inf that a grief.howl from sich gegeben hätten und dreimal über eine Statue refl give-ppp have-sbjv.pst and three.times above a statue des verstorbenen Präsidenten geflogen seien.¹⁹⁹ the-gen deceased-gen president-gen fly-ppp be-sbjv.prs ‘Reportedly, three wild geese have been observed in the northwest of Pyongyang that were howling in grief and that were flying three times over the statue of the deceased president.’

(263)

Der Thüringer Verfassungsschutz soll laut Medien the Thuringian protection.of.constitution shall according Media den Mitgliedern des Neonazi-Trios in Zwickau vor the members the-gen Neo-Nazi-Trio-gen in Zwickau before elf Jahren 2000 Mark zum Kauf gefälschter eleven years 2000 Mark for.the purchase falsified-gen Pässe gegeben haben.²⁰⁰ passport-gen give-ppp have-inf ‘According to the media, the Thuringian Office for the Protection of the Constitution has supported the members of the Neo-Nazi-Trio in Zwickau eleven years ago with 2000 Mark for the purchase of falsified passports.’

Example (262) is of particular interest, as it exhibits a remarkable interaction of a reportative modal modifier and other grammatical means to qualify a proposition as a claim that has been made by somebody other than the speaker, such as

198 http://www.orf.at/stories/2042591/2042574/, 16th February 2011. 199 http://www.orf.at/stories/2096125/2096140/, accessed on 22th December 2011. 200 http://derstandard.at/1324170159908/Verfassungsschutz-soll-Nazi-Trio-finanziert-haben 19/12/2011, accessed on 19th December 2011.

2.2 Case studies |

141

the subjunctive of the present seien ‘be-sbjv.prs’ and the subjunctive of the past hätten ‘have-sbjv.pst’. On closer inspection, it turns out that this configuration involves some sort of ‘sequence of modality effect’ in analogy to the sequence of tense effect in languages such as English or French. The proposition of the superordinate clause is identified as a claim made by somebody else. According to this, the two embedded conjoined relative clauses appear to require a similar specification. For some reason, the author employs the subjunctive of the past hätten in the first conjunct and the subjunctive of the present seien in the second. If both finite verbs were specified as indicative of the present, the pattern would be significantly less acceptable. It is important to stress that reportative sollen is not restricted to the modification of stative predicates, it is also documented with predicates that denote a process, such as (264). (264)

Schnaps und Wein mit dem Konterfei Adolf Hitlers auf dem schnapps and whine with the portrait Adolf Hitler-gen on the Etikett: Dafür soll ein Vorarlberger im Internet label therefore soll a Vorarlbergian in.the Internet werben.²⁰¹ advertise-inf ‘Reportedly, a Vorarlberger advertised schnapps and wine with a label exhibiting a portrait of Adolf Hitler on the Internet.’

Just as with reportative wollen, the reportative use of sollen is not subject to the CoDeC to the same extent as epistemic modal verbs are, as will be pointed out in more detail in Chapter 6. In opposition to epistemic modal verbs, the speaker can agree or disagree with the embedded proposition which is labelled as a claim of another referent. He may even know whether it is true or false. Similar observations have been made by Öhlschläger (1989: 235), Ehrich (2001: 157), Colomo (2011: 241), Faller (2011: 4, 2012: 289). As has been already illustrated in Section 2.2.1, epistemic modal verbs can be characterised by means of the environments from which they are excluded. As shown by Reis (2001: 294, 296), sollen occurs more readily in contexts in which epistemic modal verbs are significantly less acceptable, such as questions or nonfinite environments. However, reportative sollen turns out to be less flexible in its behaviour than reportative wollen. This may be due to the fact that the attitude holder is provided as an implicit argument of a raising verb in the case of sollen, whereas, in the case of wollen, the attitude holder is encoded in a more salient manner: It is realised as a subject argument.

201 http://vorarlberg.orf.at/news/stories/2512078/, accessed on 6th December 2011.

142 | 2 Modal verbs: A class struggle

Interestingly, sollen appears to involve a concessive resonance in some cases, just as with concessive epistemic mögen in Section 2.2.7.7, and with the instances of wollen (cf. 192–194) discussed in Section 2.2.3.8. The author of the utterance (265) makes the concession that the embedded proposition can be considered possible. (265)

Dass Renyi bei einem Blatt arbeitet, das mit dem „Falter” that Renyi at a paper works that with the Falter vergleichbar ist und schon seit 1991, verstärkt dann ab comparable is and already since 1991 intensified then after 1993, gegen „Fidesz” kampagnisiert – soll sein, beweist aber 1993 against Fidesz campaigns shall be-inf proves but eigentlich nur, dass es um die Pressefreiheit in Ungarn so actually only that it about the freedom.of.press in Hungary so schlecht nicht bestellt sein kann.²⁰² bad neg tilled be-inf can ‘That Renyi works for a newspaper that could be compared to the “Falter” and that campaigns against “Fidesz” since 1991 and from 1993 onward in a more intensive manner may be right. But even if so, this just proves that the freedom of the press cannot be severely endangered in Hungary.’

Data such as (265) support the hypothesis that concessive semantics could be related to volitional meaning. All of the epistemic modal verbs that allow for a concessive interpretation mögen, wollen and sollen involve volitional semantics to some extent or another. There are other Germanic languages in which reportative modal interpretations of the counterpart of sollen can be found, e.g. Danish skal ( Palmer (1986: 72)) and Norwegian skulle (cf. Eide (2005) and Hetland and Vater (2008: 96)). Interestingly, these languages lack reportative interpretations of the counterpart of wollen, from which the interpretation of sollen is derived. 2.2.6.5 Raising infinitives with clause modification: epistemic Apart from its reportative interpretation that refers to a claim attributed to another person, sollen can express an assumption made by the speaker if it bears subjunctive of the past morphology. This has been already observed by Becker (1836: 181), Bech (1949: 16), Glas (1984: 104), Öhlschläger (1989: 236 Fn. 223), Scholz (1991: 275 Fn. 114) and Fritz (1997: 107). Whenever sollte selects a predicate that refers to a state that cannot be changed (cf. 266) or a predication that refers to an event in the past (cf. 267), it is

202 Die Presse, 29/12/2011.

2.2 Case studies |

143

restricted to an epistemic interpretation. Once again, a circumstantial reading is not available in such configurations. (266)

Das Software-Problem sollte aber mittlerweile behoben the software-problem shall-sbj.pst but meanwhile resolve-ppp sein, wie SBB-Sprecher Jean-Louis Scherz sagte.²⁰³ be-inf as SBB-spokesman Jean-Louis Scherz said ‘The software problem should be resolved by now, as indicated by the SBB-spokesman Jean Louis Scherz.’

(267)

Der jüngste Erfolg sollte auch das Nervenkostüm the youngest success shall-sbj.pst also the nerve.costume seiner Mannschaft stabilisiert haben.²⁰⁴ his-gen team-gen stabilise-ppp have-inf ‘The last success should have also rendered his team a bigger self assurance.’

(268)

Selbstbewusstsein sollte eigentlich auch bei der SG self.confidence shall-sbj.pst actually also by the SG Unnertal in Massen vorhanden sein.²⁰⁵ Unnertal in masses present be-inf ‘SG Unnertal should actually have plenty of self confidence.’

(269)

Denn in vierzehn Tagen sollte es mit Kälte und Frost since in fourteen days shall-sbj.pst it with cold and freeze theoretisch vorbei sein.²⁰⁶ theoretically over have-inf ‘Because, theoretically, in these fourteen days, the cold and the freeze should have already gone .’

Interestingly, epistemic sollte frequently occurs in the environment of two specific speech act adverbs: eigentlich ‘actually’, (cf. 268) and theoretisch ‘theoretically’, (cf. 269). These preferences could be helpful to identify the exact interpretation of epistemic sollte. First of all, the question of what conditions cause the shift of the deictic centre deserves closer attention. Whereas reportative sollen identifies the deictic centre with an implicit argument, epistemic sollen links the deictic centre to the speaker referent. Obviously, a similar shift takes place with circumstantial sollen. Bearing subjunctive of the past morphology, the source of volition is more likely to be the

203 204 205 206

DeReKo: A09/DEZ.04148 St. Galler Tagblatt, 14/12/2009. DeReKo: BRZ09/MAI.07459 Braunschweiger Zeitung, 16/05/2009. DeReKo: RHZ99/AUG. 20012 Rhein-Zeitung, 28/08/1999. DeReKo: O98/MAR.21556 Neue Kronen-Zeitung, 05/03/1998.

144 | 2 Modal verbs: A class struggle

speaker in the case of sollte. In example (270) the referent to whom the volition is attributed is the speaker, identical to Lutz Greiner. Even if the context is changed, it is more difficult to obtain an interpretation in which the source of volition is identified with another referent. (270)

Auch Lutz Greiner würde vor allem stören, wenn dort also Lutz Greiner would above all annoy if there etwas gebaut würde. „Man sollte nicht die ganze something built would one should neg the whole Landschaft zubetonieren. [. . . ]”²⁰⁷ landscape to.concrete-inf ‘ Lutz Greiner would also be annoyed if something was built there. “One should not cover the whole landscape with concrete” ’

Moreover, epistemic sollte exhibits a semantic peculiarity. As observed by Copley (2006: 11), epistemic should in English is only acceptable if it refers to remote evidence. In a scenario in which the speaker sees that the lights are turned on in the guest’s house uttering the sentence (271) would not be appropriate, whereas must would be. Copley’s observation can neatly be extended to German epistemic sollte, as is illustrated in (272). (271)

Our guests should be home by now.

(272)

Unsere Gäste sollten mittlerweile schon zuhause sein. Our guests should meanwhile already at.home be-inf ‘Our guests should be home by now.’

As Copley (2006: 5) has further pointed out, the speaker can be aware that the proposition embedded by should is false. Accordingly, the speaker can resume the discourse in a context as in example (271): ‘. . . but they aren’t’. Once again, this equally holds for sollte in German. A similar situation obtains for the other epistemic modal verbs that may bear subjunctive of the past morphology, könnte ‘can-sbjv.pst’ (cf. Section 2.2.1.5) and müsste ‘must-sbjv.pst’ (cf. Section 2.2.2.5). In contrast, the instances above are all subject to the CoDeC. Employing sollte, the speaker indicates that the embedded proposition is not part of his knowledge. There are two essential properties of sollte. First, just as with the other subjunctive epistemic modal verbs könnte and müsste, sollte indicates that at least one of the premisses on which the conclusion is based is non-verified or even counterfactual (cf. Section 2.2.1.5 and 2.2.2.5). This is seen with the most acceptable examples, which involve an implicit antecedent of a conditional (cf. 273iii, 274i, 274ii, 275ii).

207 DeReKo: BRZ09/JAN.09361 Braunschweiger Zeitung, 23/01/2009.

2.2 Case studies |

145

Es sollte gerade regnen. it should now rain-inf

(273)

‘It should be raining now.’

i.

# I hear some sound of patter on the roof.

ii.

# I see people with umbrellas coming in.

iii. iv.

If the weather forecast was right ? I saw some heavy clouds heading towards our house a couple of minutes ago.

Wir sollten in einer Stunde zuhause sein we shoud in a hour home be-inf

(274)

‘We should be home in an hour.’

i.

If we are going to take the next train.

ii.

If there is not much traffic ahead of us

Sebastian sollte den Zug versäumt haben. Sebastian should the train miss-inf

(275)

‘Peter should have missed the train.’

i. ii.

# He is not in his office. Unless he found his wallet which I hid on purpose.

[Es klingelt] Das sollte Zwentibold sein. it rings this should Zwentibold be-inf

(276)

[The door bell rings] ‘This should be Zwentibold.’

i. ii.

I saw him entering the staircase a minute ago. # I didn’t expect him. May be he left some thing here.

This indicates that epistemic sollte retains much of its counterfactual semantics indicated by its morphology. As the most prototypical examples of sollte involve an implicit conditional antecedent, one could consider epistemic sollte as truncated conditional. These antecedents often point at the stereotypical course of how things go. Likewise, similar antecedents can be found for the corpus examples discussed above. Example (266) could be complemented with an antecedent like if everything went right, examples (267) and with an antecedent like if the players are not too unassured, and finally (269) can be complemented with an antecedent like if the model for the weather forecast was calculated correctly. The second requirement concerns the epistemic modal base or the type of premises on which the conclusion is drawn. Copley (2006: 11) pointed out that the evidence on which the epistemic judgement for sollte is made has to be remote evidence. Hinterwimmer’s (2014) observation according to which sollen asserts a previous intentional act seems to be related. Putting these two observations

146 | 2 Modal verbs: A class struggle

together one could assume that epistemic sollte selects a modal base which only contains propositions which were added some critical time before Utterance Time. This analysis is supported by the behaviour of sollte in the contexts illustrated in (273)–(ex:SollteZwentiboldSein). First of all, it is demonstrated here that an epistemic interpretation is only available in example (273) if all premisses on which the epistemic conclusion is based were known before Utterance Time: in a context like (273iii), example (273) is acceptable. In contexts (273i) and (273ii), where the premisses are added to the knowledge of the epistemic agent during utterance time, epistemic sollte is not acceptable. It is not so much what the senses (eyes, ears) report to you directly but the premisses have to be part of the episodic knowledge. Another important characteristic of epistemic sollte is that the epistemic agent must have expections that the prejacent would happen and that these expectations were already part of his episodic knowledge before utterance time, as demonstrated in (276): in the context (276ii) in which it hasn’t already been an expectation before the epistemic evaulation that Zwentibold would come, epistemic sollte is not available. Summing up, epistemic sollte seems to be the epistemic modal verb whose modal base exhibits the highest degree of episodic knowledge. It is not compatible with any visual or auditive evidence which does not convey propositional content such as written or spoken language and which is accessible at Utterance Time. To some extent, sollte seems to behave in a manner parallel to dürfte: Both can only be interpreted in an epistemic way if they bear subjunctive morphology. Following the observations made by Bech (1949), one could argue that this could be due to the fact that both verbs involve a volitional resonance in their circumstantial interpretation. However, sollte and dürfte differ with respect to a crucial property. As has been illustrated in Section 2.2.5, epistemic dürfte behaves very opaquely. As a consequence, it cannot be compositionally derived from its circumstantial counterpart. Thus, it has to be considered as an independent lexical item. Some authors, such as Diewald (1999: 202) claim that, in these instances of sollte, the speaker does not make an epistemic evaluation of the embedded proposition. Her conclusion is based on the evidence that sollte cannot be substituted with epistemic dürfte without affecting the interpretation of the entire sentence. It is doubtful whether the difference in the interpretation is really the right evidence to support her claim. As shown above, there are many properties of sollte which are much in favour of an analysis of sollte as an epistemic modifier and that would be left unaccounted for otherwise. It is not clear how Diewald would explain that sollte is subject to the CoDeC and becomes epistemic just in the same environments as all of the other remaining epistemic modal verbs.

2.2 Case studies |

147

2.2.7 mögen As has been pointed out by Öhlschläger (1989: 176), mögen represents a special case in semantic respects. It involves a whole range of usages which are intricate to capture. First of all, it can be used as a transitive verb, which is the most frequent pattern. In this use it embeds finite dass-clauses and it licences wennclauses. Apart from that, it can be used as a volitional control verb, though in most varieties only as a negative polar item. Furthermore, a circumstantial raising pattern is hard to find for mögen, there are at best some highly idiosyncratic instances that could be considered raising verbs. Then, it always exhibits a concessive resonance whenever employed as an epistemic modal verb, and its “purely” epistemic possibility reading has almost been lost in the last centuries. Moreover, it has a defective morphological paradigm, as its synthetic subjunctive of the past möchte has emancipated and has become an independent lexical item, to be discussed in Section 2.2.8. Finally, it merits closer attention that mögen is by far the least frequent item among the traditional six modal verbs in contemporary spoken language, as Ruoff (1981) has pointed out and as was also shown in Section 2.1.4. The peculiarities of mögen can be explained in terms of the diachronic development of the entire group of modal verbs. As has been illustrated by Bech (1951: 23), Fritz (1997: 9) and Diewald (1999: 392), mögen used to be the default possibility modal verb until the Early New High German period, when it came gradually to be replaced with the new possibility modal verb können. As a consequence, mögen acquired a semantic component: an emotive/volitional feature. As it seems, the uses with an infinitive erode and the more lexical uses with an accusative NP or with finite complement clauses become more important. So it would not be too surprising if mögen dropped out of group of modal verbs in the course of the next centuries. 2.2.7.1 Transitive uses Among the traditional six modal verbs, mögen is the one with the most evident transitive use. It indicates an affection between an animate subject referent and the referent realised as the direct object in an active clause, as has been demonstrated by Becker (1836: 180), Öhlschläger (1989: 69), Diewald (1999: 192) and Eisenberg (2004: 96). Just like any other transitive verb, mögen can be found in passivisations, as is illustrated in (277)–(278). (277)

Als verwöhnte Zicke wird sie später von niemandem as fastidious bitch pass.aux she later by nobody gemocht.²⁰⁸ like-ppp

148 | 2 Modal verbs: A class struggle

‘Being a fastidious bitch, she will not attract much affection later on.’

(278)

Als Schriftsteller wird er [Henry de Montherlant] von as writer pas.aux he-nom Henry de Montherlant vielen nicht gemocht, weil er [. . . ] am übersteigerten by many neg like-prt.pas since he at overreaching Männlichkeitsgefühl, dem sogenannten Machismo, litt.²⁰⁹ feeling.of.masculinity the so-called Machismo suffered ‘As a writer, Henry de Montherlant is not very popular because he suffered from a overreaching feeling of masculinity, the so-called Machismo.’

As Becker (1836: 180) has pointed out, this pattern already occurs in Luther’s works in the early 16th century. In comparison to the remaining five traditional modal verbs, mögen is the one that occurs most often in passives. Nevertheless, it exhibits a morphological anomaly. Mögen has lost its synthetic subjunctive of the past. Having developed independent semantics, its genuine form möchte is always construed with a volitional interpretation (cf. 280). It is not synonymous with the analytic subjunctive of the past in (cf. 281) anymore. Accordingly, the analytic pattern is the only one that is acceptable in counterfactual environments. (279)

Chihiro mag Natto. Chihiro likes Natto ‘Chihiro likes Natto.’

(280)

# Lola möchte

Natto, wenn Sie keine Katze wäre. Lola like-sbjv.pst Natto if she no cat be-sbjv.pst

‘Lola likes to have some Natto, if she was not a cat.’

(281)

Lola würde Natto mögen, wenn Sie keine Katze wäre. Lola would Natto like-inf if she no cat be-sbjv.pst ‘Lola would like Natto, if she was not a cat.’

By contrast, möchte cannot be interpreted as mögen in the scope of a counterfactual operator anymore; similar observations have been made by Lötscher (1991: 338, 354). 2.2.7.2 dass-clauses Aside from its transitive use, emotive mögen can occur with finite dass-clause complements. The emotive use of mögen presupposes that the proposition expressed by the dass-clause is factual. Interestingly, the majority of occurrences of emotive

208 DeReKo: M06/FEB.15690 Mannheimer Morgen, 24/02/2006. 209 DeReKo: O97/APR.41025 Neue Kronen-Zeitung, 20/04/1997.

2.2 Case studies |

149

mögen found in the DeReKo corpus originates from newspapers from West Central and North Western regions.²¹⁰ There are almost no instances from Switzerland or Austria. As most of the occurrences involve 1st person subjects, it can be concluded that this pattern is primarily characteristic of the spoken language, as is illustrated in (282). However, there are rare cases in which the subject bears the feature 3rd person, as in (283). The situation for emotive mögen in the scope of negation is similar. Cases with 3rd person subjects, which are exemplified in examples (284)–(285) are rather rare. (282)

Ich mag, dass im Frühling alles blüht.²¹¹ I like that in.the Spring everything blossoms ‘I like the fact that in Spring everything is blossoming.’

(283)

Monika Pohl und Susanne Schnaidt mögen, dass Schmuckstücke Monika Pohl and Susanne Schnaidt like that trinkets flexibel sind, wie die Ohrhänger aus unbehandeltem, flexible are like the earring of untreated gewachsenen Türkis, die auch als Kreolen getragen grown Turkey.stone rel.prn also as Creoles wear-ppp werden können.²¹² pass.aux-inf can ‘Monika Pohl and Susanne Schnaid like the fact that trinkets are flexible such as the earring made out of untreated naturally grown Turkey stone which can also be worn as Creole.’

(284)

Michael Schumacher mag nicht, dass ein anderer dafür Michael Schumacher likes neg that a other therefore bezahlt wird, so auszusehen wie er.²¹³ pay-ppp pass.aux so to.outlook as he ‘Michael Schumacher does not like the fact that there is somebody who is paid just to look like him.’

(285)

Die Schlange war wohl ähnlich nervös wie heute, mochte the snake was maybe similarly nervous as today like nicht, dass ein Mitschüler sie am Schwanzende packte.²¹⁴ neg that a class.mate she at.the tail.end grasped ‘The snake had been obviously as nervous as it was today and did not like it that a class mate grasped it by its tail.’

210 Survey carried out 21st May 2012. 211 DeReKo:BRZ07/JUN.01103 Braunschweiger Zeitung, 14/06/2007. 212 DeReKo: M08/MAI.35471 Mannheimer Morgen, 09/05/2008. 213 DeReKo: N00/MAR.14463 Salzburger Nachrichten, 31/03/2000. 214 DeReKo: BRZ06/AUG.04657 Braunschweiger Zeitung, 10/08/2006.

150 | 2 Modal verbs: A class struggle

Crucially, in all of examples with emotive mögen the embedded proposition is presupposed to be factive (cf. 282–285). In order to felicitously utter example (282), it is required that everything be usually blossoming. These context conditions are very different from those of circumstantial modality. As Ziegeler (2006) has pointed out, in the most canonical case the proposition embedded by a circumstantial modal operator is not true at the Time of Utterance. Apart from the emotive use of mögen, there is another one that can embed a finite dass-clause with a volitional interpretation. In contrast to the emotive pattern, the proposition of the embedded clause is not presupposed to be factive in the volitional interpretation. In example (286), it is not granted that Lang has yet been considered an intellectual. The counterfactual nature of the dass-clause in this example becomes obvious by shifting the attention to the consecutive clause: The use of the subjunctive of the past in the consecutive clause indicates that in the actual world Lang has not yet been called an intellectual. In opposition to its emotive use, mögen in examples (286)–(288) aims at the realisation of the embedded predication, rather than evaluating an established fact. As can be seen, in most varieties the volitional use of mögen requires the presence of a negative operator in the clause. At least in Upper East German varieties, volitional mögen with dass-clauses is even found without negation, as exemplified in (288). (286)

Lang mag nicht, dass man ihn einen Intellektuellen nennt. „Das Lang likes neg that one him a intellectual calls that wäre eine Übertreibung, zu sehr Schublade”; wehrt er ab is-sbjv.pst a overstatement too much drawer wards he off und fordert, dass man weg kommt von der Unsitte der and demands that one away gets from the bad.habit the-gen Show in der Politik.²¹⁵ show-gen in the politics ‘Lang does not like that he is considered as intellectual. “This would be an overstatement, too much pigeonholing” he objects and demands that one should refrain from the bad habit of making too much show in politics.’

(287)

»Wirklich, du bist gar nicht findig, Roswitha. Und ich mag Truly you are intn neg resourceful Roswitha and I like nicht, daß du dich erkältest, und alles um nichts.«²¹⁶ neg that you refl catch.a.cold and everything for nothing. ‘Truly, you are not very resourceful, Roswitha. And I do not want that you catch a cold and all that would be for nothing.’

215 DeReKo: RHZ04/JUN.08194 Rhein-Zeitung, 08/06/2004. 216 Theodor Fontane, Effi Briest Chapter 21. (1896).

2.2 Case studies |

(288)

151

Verdammt, ich mag, dass du wieder in Wien bist! Komm vorbei Damn I like that you again in Vienna are come along – du bist bei uns herzlichst willkommen!!²¹⁷ you are at us cordially welcome ‘I wish you were back in Vienna! Come along – you are always cordially welcome!’

As can be seen, the negative polarity of volitional mögen with dass-clause is tied to its particular volitional semantics. As will be shown in more detail below, volitional mögen with bare infinitive complement exhibits the same preference for negative environments. 2.2.7.3 wenn-clauses Among the traditional six modal verbs, mögen is the only one that occurs with wenn-clauses. As Fabricius-Hansen (1980: 162) and Kaiaty (2010: 305) indicate, determining the syntactic status of these clauses is no trivial matter. They involve properties of both complement clauses and adverbial clauses. On the one hand, Fabricius-Hansen (1980: 164) and Kaiaty (2010: 288) illustrate that this precise type of wenn-clause is restricted to a particular class of matrix predicates. On the other hand, Fabricius-Hansen (1980: 163) and Kaiaty (2010: 289) show that these wenn-clauses require the presence of a correlate that saturates the relevant argument position of the matrix predicate in the canonical case. Being a preference predicate, emotive mögen is expected to occur with wenn-clauses, as Kaiaty (2010: 293) argues. It is indeed found in the DeReKo corpus. (289)

ich kann nicht sagen, daß sie es nicht mochte, wenn Oskar I can neg say-inf that she it-cor neg liked if Oskar ihr darunter saß.²¹⁸ her underneath sat ‘I cannot deny that she did not liked it when Oskar sat underneath her.’

Fabricius-Hansen (1980: 185) and Kaiaty (2010: 305) argue that the type of wennclause under discussion has to be considered an adverbial clause. The position of the theme argument provided by mögen is typically saturated by the correlate es, as in example (289). The function of the wenn-clause, in turn, is to make the state of affairs the correlate relates to explicit. However, there are also instances of mögen with wenn-clauses that do not overtly select a correlate, cf. (290) and (291).

217 de-de.facebook.com/MeinOlliSchulz/posts/10150584251049316, accessed on 21st May 2012. 218 DeReKo: MK1/LGB.00000 Grass, Günter: Die Blechtrommel, (1962).

152 | 2 Modal verbs: A class struggle

(290)

Menotti hat nie gemocht, wenn einer hohe Flanken Menotti had never liked when someone high crosses schlug.²¹⁹ made ‘Menotti never liked it when someone played high crosses.’

(291)

Er mag, wenn es einem scheinbar Schwachen gelingt, sich he likes when it a putatively weak.one manages refl gegen Stärkere durchzusetzen.²²⁰ agains stronger prevail ‘He likes it when a putatively weak person manages to prevail against a stronger one.’

Essentially, there is one major difference between wenn-clauses and dass-clauses that are embedded under emotive mögen: While a dass-clause always refers to a factive proposition, the proposition expressed by a wenn-clause need not be factive, as Kaiaty (2010: 295) has pointed out. Rather, it receives a future oriented or conditional interpretation. Under very restricted conditions, volitional verbs such as wollen can also select correlates that refer to a proposition which is made explicit by a wenn-clause. (292)

Er will die inhaltliche Diskussion. Was er nicht will, ist, he wants the content-adj discussion what he neg wants is wenn daraus sozusagen Seilschaften gebildet oder when out.of.it so.to.speak rope.teams form-ppp or Grabenkämpfe gemacht werden.²²¹ trench.warfare make-ppp pass.aux-inf ‘He wants a discussion about the content. What he does not want is if this turns into insider relationships or partisanship.’

In example (292), will occurs in the scope of negation and it is part of a wh-cleft. Crucially, in all of the examples discussed above (289)–(291), mögen cannot be replaced with wollen. 2.2.7.4 Control infinitives with event modification As has been pointed out by Welke (1965: 115), Öhlschläger (1989: 179) and Weinrich (1993: 307), mögen can be combined with a bare infinitive complement, yielding a volitional interpretation. Further, Bech (1949: 21), Welke (1965: 115), Weinrich (1993: 307), Diewald (1999: 288, 315–317) notice that the volitional interpretation

219 DeReKo: E98/JUN.15928 Zürcher Tagesanzeiger, 29/06/1998. 220 DeReKo: HMP08/DEZ.00738 Hamburger Morgenpost, 08/12/2008. 221 DeReKo: RHZ07/SEP.04038 Rhein-Zeitung, 05/09/2007.

2.2 Case studies |

153

seems to be restricted to environments in which mögen occurs in the scope of a negative operator, as examples (293)–(294) indicate. However, in some varieties it can be used without a negation, as shown in (295)–(296). The precise meaning of volitional mögen is difficult to capture, but it seems to be similar to the one of volitional wollen. In most of the examples below, mögen can be substituted with wollen without drastically affecting the overall meaning. In opposition to its volitional counterparts, volitional mögen with an infinitive occurs less frequently, by far, at least in written language. (293)

Über Geld mag er nicht reden.²²² About money likes he neg talk-inf ‘He does not want to talk about money.’

(294)

Stefan Müller, Abgeordneter aus Erlangen und einflussreicher Stefan Müller depute from Erlangen and influential Chef der Jungen Union, mag die Krise der CSU boss the-gen Junge-gen Union-gen wants the crisis the-gen CSU erst gar nicht kleinreden.²²³ only intn neg play.down-inf ‘Stefan Müller, depute from Erlangen and influential boss of the Junge Union does not want to play down the crisis of the CSU.’

(295)

„Doch, er ist der erste Mann, mit dem ich richtig gerne yes he is the first man with that I truly willingly zusammenleben mag, eben weil wir uns Freiräume lassen”, together.live-inf like just because we us free.space leave sagt die schöne Habermann.²²⁴ says the beautiful Habermann ‘Oh yes, he is the first Mann with whom I really would like to live together, just because we allow us room for ourselves.’

(296)

Zoë ist ihre kleine Cousine. „Ich mag den aber anziehen”, beharrt Zoë is her small cousin I like this but put.on-inf insists sie und schlüpft ruckzuck in das Teil.²²⁵ she and like.a.shot in the piece ‘Zoë is her small cousin “But I want to put on this” she insists and slips into the piece.’

It merits closer attention that the behaviour of volitional mögen with an infinitive reflects the one of its volitional counterpart with dass-clauses: In Standard

222 223 224 225

DeReKo: HMP09/MAR.02623 Hamburger Morgenpost, 25/03/2009. DeReKo: NUN09/JAN.01879 Nürnberger Nachrichten, 04/04/2008. DeReKo: HMP09/AUG.01928 Hamburger Morgenpost, 19/08/2009. DeReKo: RHZ06/NOV.17231 Rhein-Zeitung, 18/11/2006.

154 | 2 Modal verbs: A class struggle

German, they are more readily acceptable if they occur in the scope of negation. This indicates that the negative polarity appears to be tied to the volitional semantics of mögen. Interestingly, negative polarity is a behaviour that occurs fairly frequently with so-called modal verbs. As has been shown by Fritz (1997: 54), there are at least three modal verbs in German that displayed a negative polar behaviour over the course of the last centuries. Volitional mögen and brauchen, which will be discussed in Section 2.2.9, can still be considered negative polarity items in contemporary Standard German. Moreover, Bech (1951: 14) observes that dürfen was a negative polarity item until the course of the 16th century. In addition to these instances, there is still a raising pattern of wollen that is negative polar, as is illustrated in Section 2.2.3.5. Crucially, mögen with a bare infinitive complement ceased to express any type of circumstantial possibility, like it did until the Early New High German period: In Contemporary Standard German, it cannot refer to a physical ability or to a practical possibility anymore. Bech (1951: 23), Fritz (1997: 9) and Diewald (1999: 392) argue that this drastic shift in meaning was due to the rise of another possibility modal verb können, which was still used rather infrequently in the Middle High German period. As soon as mögen had acquired the volitional feature, all of the circumstantial possibility readings presumably ceased to exist. Finally, Welke (1965: 114) has pointed out that mögen can be part of some idiomatic expressions, such as leiden mögen ‘like’. 2.2.7.5 Control directionals with event modification Just like all of the other circumstantial modal verbs described in this section, volitional mögen can select verbless directional phrases, as exemplified in (297)–(298). The semantic behaviour is parallel to the one of mögen with a bare infinitive complement. (297)

„Mein Sohn Julian mag nicht mehr in den Kindergarten”, sagt my son Julian likes neg more in the kindergarten says etwa Nina Islitzer²²⁶ for.instance Nina Islitzer ‘ “My son Julian does not like to go to the kindergarten any more” says Nina Islitzer for instance.’

(298)

Wer dann nicht mehr zurück in die Stadt mag: Es gibt who then neg more back in the town wants it gives Hotelzimmer im „Eichbaum”.²²⁷ hotel.rooms in.the Eichbaum

226 DeReKo: NON09/SEP.18165 Niederösterreichische Nachrichten, 29/09/2009.

2.2 Case studies |

155

‘Whoever does not want to return to the city, there are rooms available at the “Eichbaum” ’

Similarly to the other volitional uses of mögen with a bare infinitive complement and with dass-clause, mögen with a verbless directional phrase is restricted to negative polarity contexts in Standard German. In some varieties, mögen with a directional phrase can occur in positive environments as well though, e.g. in East Upper German. 2.2.7.6 Raising infinitives with event modification In Contemporary German, raising patterns of mögen with a circumstantial interpretation appear to have died out. However, there are some archaic uses that are candidates for a raising analysis. Schoetensack (1856: 294) observes that mögen can refer to a volition attributed to a third party, as in is illustrated in (299). (299)

Er möge hereinkommen.²²⁸ he likes-sbjv.prs enter-inf ‘May he enter!/I allow/want him to enter.’

Crucially, there is no thematic relation between mögen and its syntactic subject in example (299). Thus, mögen does not carry a subject argument in the example given above. Rather, möge expresses that the embedded proposition is consistent with the wishes of the speaker. In this respect, it is reminiscent of the raising use of wollen discussed in Section 2.2.3.5. 2.2.7.7 Raising infinitives with clause modification The epistemic uses of mögen are fairly peculiar and very hard to analyse. Moreover, Öhlschläger (1989: 187) observes that they occur very rarely, and that they carry an interpretation that is very different from their circumstantial counterparts. As can be seen, a couple of different patterns have to be distinguished. As has been pointed out by Becker (1836: 180), Bech (1949: 23), Welke (1965: 110), Allard (1975: 88),Öhlschläger (1989: 187 Fn. 121), Fritz (1991: 48), Weinrich (1993: 314) and Diewald (1999: 236), epistemic mögen usually conveys a concessive resonance and behaves in a marked way. Like all of the remaining modal verbs, mögen is restricted to an epistemic interpretation in two particular contexts: Whenever it embeds a predication consisting of an identified individual and a predicate that denotes a state that cannot be

227 DeReKo:HMP06/JUL.02708 Hamburger Morgenpost, 27/07/2006. 228 Example as quoted by Schoetensack (1856: 294).

156 | 2 Modal verbs: A class struggle

changed (cf. 300–302), any circumstantial interpretation is excluded. Likewise, mögen can only be interpreted in an epistemic way whenever it embeds a predication that refers to an event in the past (cf. 303–306). (300)

Mit 8 Milliarden Euro Kosten mag Nabucco teuer sein – sie with 8 billion Euro costs may Nabucco expensive be she wird sich bezahlt machen.²²⁹ will refl pay-ppp make-inf ‘With its costs of around 8 billion Euros, Nabucco may be expensive – nevertheless, it will pay off.’

(301)

Candye Kane mag hundert Kilo schwer sein, ist bei Gott keine Candye Kane may hundert Kilo heavy be-inf is by God no klassische Schönheit und spielt nostalgischen Blues. Dennoch classical beauty and plays nostalgic blues nevertheless verfällt man dieser Frau: Sie ist auf der Bühne, um one addicts this woman she is on the stage in.order musikalisch zu heilen.²³⁰ musically to heal-inf ‘Candye Kane may weigh 100 kilos, she is by no means a classical beauty, and she plays nostalgic blues. Nevertheless, one addicts to this woman: she is on stage to heal with music.’

(302)

Das Mädchen mag erst drei Jahre alt sein. Doch aufmerksamer the girl may just three years old be-inf but attentive-comp war wohl noch nie ein Buspassagier. Kaum wird was maybe still never a bus.passenger as.soon.as namely nämlich die Haltestelle »Singenberg« angesagt, da pass.aux the stop Singenberg announce-ppp there beginnt das Mädchen auch prompt zu singen: Singenberg, begins the girl also immediately to sing-inf Singenberg, Singenberg, Singenberg . . . – angepasst der Melodie von Hänschen Singenberg, Singenberg adapted the melody of Hänschen klein.²³¹ klein ‘The girl may be just three years old. But no passenger of the bus has ever been more attentive. As soon as the stop »Singenberg« is announced, the girls immediately starts to sing: Singenberg, Singenberg, Singenberg . . . –’ adapting the melody of Hänschen klein.’

229 DeReKo: NON09/JUL.05420 Niederösterreichische Nachrichten, 13/07/2009. 230 DeReKo: A07/OKT.07573 St. Galler Tagblatt, 17/10/2007. 231 DeReKo: A00/JAN.03266 St. Galler Tagblatt, 15/01/2000.

2.2 Case studies |

(303)

157

Feminismus mag in der modernen Gesellschaft wirklich Feminismus may in the modern society indeed etwas Wichtiges bewirkt haben, aber braucht man diese something important cause-ppp have-inf but needs one this Bewegung auch in der Kirche?²³² movement also in the church? ‘Even if Feminism may have caused important changes in the modern society, is it a necessary movement also in the church?’

(304)

Dieser Satz mag im Affekt gefallen sein, doch empfinde this sentence may in.the affect fall-ppp be-inf but perceive ich ihn als symptomatisch für unsere Zeit.²³³ I it as symptomatic for our time ‘This sentence may have been uttered in the heat of the moment, nevertheless I consider it as symptomatic for our time.’

(305)

Das ist psychologisch einfach zu erklären. Diese Kreditvorlage, that is psychologically simply to explain this credit.approval so gut sie gemeint sein mag, kommt zum falschen so well it mean- be-inf may comes at.the false Zeitpunkt.²³⁴ moment ‘This can be accounted for in terms of psychology. This credit approval arrives in the wrong moment, even if it was well meant.’

(306)

Sein Äußeres mag sich verändert haben, seinen Idealen ist his appearence may refl change-ppp have-inf his ideals is Carlo Acquistapace treu geblieben.²³⁵ Carlo Acquistapace loyal stay-ppp ‘His appearance may have changed, but as regards his ideals, Carlo Acquistapace remained true to them.’

The semantic contribution of concessive epistemic mögen is intricate. A speaker who utters a proposition p in the scope of a concessive epistemic mögen implicitly makes a whole range of statements about p: First of all, he indicates that there is another referent who believes p to be true. Secondly, the speaker himself did not believe p to be true. Thirdly, by uttering the sentence mögen(p), he conveys that he changed his mind and considers p to be possible and consistent with his own

232 233 234 235

DeReKo: RHZ09/MAI.08761 Rhein-Zeitung, 11/05/2009. DeReKo: HAZ08/AUG.05647 Hannoversche Allgemeine, 28/08/2008. DeReKo: 00/AUG.51643 St. Galler Tagblatt, 04/08/2000. DeReKo: RHZ09/AUG.18314 Rhein-Zeitung, 22/08/2009.

158 | 2 Modal verbs: A class struggle

knowledge. Fourthly, the speaker stresses that the fact that p is possibly true is irrelevant to the surrounding discourse. At this point, it is not evident how these different aspects of the interpretation of concessive epistemic mögen can be described in a more systematic and principled manner. Thus, the concessive epistemic use of mögen appears to make reference to an external referent to which an epistemic state is attributed. In this property, it very much resembles the quotative use of sollen, as well as the raising use of wollen, which ascribes a volition to a third referent, as was shown in Section 2.2.3.5. Interestingly, both sollen and wollen seem to occur with a concessive epistemic interpretation as well, as pointed out in the relevant sections. It is not evident to what extent this concessive epistemic use is subject to the CoDeC. Assuming that the deictic centre is identified with the speaker, the modified proposition should not be part of the knowledge of the speaker. It is not completely ruled out that the author of the utterance in (306) is actually aware that Acquistapace’s appearance has changed. Accordingly, the concessive use would drastically differ from the remaining epistemic modal verbs. For similar reasons, some authors, such as Öhlschläger (1989: 187), argue that it is questionable whether concessive mögen can be considered an epistemic modal at all. However, it is not entirely clear whether the speaker in example (306) is willing to fully accept that the modified proposition (Acquistapace’s appearence has changed) is true. Certainly, this proposition is not part of the strongest conviction of the speaker. As a consequence, concessive epistemic mögen could act as an existential quantifier over possible worlds: There is at least one world in the modal base in which the proposition is true, and this world is a world that is not very favourable for the speaker; but it is favourable for a third party. This ranking could be expressed by an ordering source, in the manner of Kratzer (1981) and Kratzer (1991). Accordingly, concessive epistemic mögen could be considered as a specialised possibility verb. Alternatively, one could apply the CoDeC to the external referent. In that case, the speaker would state that he would not attribute the modified proposition to the knowledge of the referent. Accordingly, he would claim that the external referent cannot really know that p is the case. It merits closer attention that the English counterpart of mögen can occasionally be interpreted with a concessive epistemic interpretation. This was already observed by Leech (1971: 69), who gives the following example: (307)

She may not be pretty but at least she knows her job.²³⁶

236 Example as quoted in Leech (1971: 69).

2.2 Case studies |

159

Apart from the concessive epistemic interpretation, mögen occurs with an interpretation in which it conveys a more neutral stance. Welke (1965: 112), Allard (1975: 89) and Diewald (1999: 236) illustrate that it can sometimes be interpreted with a pure epistemic possibility interpretation. As Welke (1965: 112) observes in his corpus study, the concessive resonance of epistemic mögen is often absent from fictional texts. In these instances, it denotes a pure assumption. In the corpus composed by Allard (1975: 89), the pure possibility interpretation prevails. The frequencies of the different interpretations are as follows: 50.64 % possibility; 18.23 % concessive; 5.58 % transition between possibility and concessive interpretation – the remaining percentage covers the non-epistemic instances of mögen. Once again, in certain environments, mögen cannot be interpreted in a circumstantial way. If it embeds a predication consisting of an identified individual and a state that cannot be changed, as shown in examples (308)–(309), mögen is restricted to an epistemic interpretation. Likewise, mögen can only be construed with an epistemic interpretation when it embeds a predication that refers to an event in the past (cf. 310). In the contexts below, the epistemic interpretation lacks the concessive resonance. Interestingly, ‘purely’ epistemic mögen is frequently found with the idiomatic expression zu tun haben ‘to have to do with, to be related to’. (308)

Wann das Kapellchen eigentlich gebaut wurde, kann niemand when the chapel actually built was can nobody mehr so genau sagen, es mag an die 100 Jahre alt sein.²³⁷ more so precise tell-inf it may on the 100 years old be-inf ‘Nobody can tell anymore when the chapel has been built, it may be about 100 years old.’

(309)

Dass Mozart auf dem Programm steht, mag gewiss mit dem that Mozart at the program stands may certainly with the fast schon inflationär gefeierten 250. Geburtstag des almost already inflationarily celebrated 250 birthday the-gen Komponisten zu tun haben.²³⁸ composer-gen to do-inf have-inf ‘It may be certainly have something to do with Mozart’s 250th birthday, which is almost celebrated in an inflationary manner.’

237 DeReKo: RHZ09/AUG.12239 Rhein-Zeitung, 15/08/2009. 238 DeReKo: NUZ06/FEB.03098 Nürnberger Zeitung, 27/02/2006.

160 | 2 Modal verbs: A class struggle

(310)

10 Uhr: Die Funkerin Margarete Wolter erwachte nach 10 o’ clock the radio.operator Margarete Wolter awoke after totenähnlichem Schlaf am anderen Morgen – es mag so dead.like sleep at.the other morning it may so gegen zehn Uhr gewesen sein.²³⁹ around ten o’ clock be-inf be-inf ‘10 o’clock: The radio operator Margarete Wolter awoke the other morning from a death-like sleep – it may have been around ten o’ clock.’

The precise semantic contribution of the epistemic use of mögen illustrated above is not obvious: Whereas Welke (1965: 110) and Zifonun (1997: 1894, 1910) conclude that epistemic mögen is synonymous with epistemic können and thus to be treated as an epistemic possibility verb with equal rights, Fritz (1997: 94), on rather intuitive grounds, assumes that epistemic mögen refers to a possibility that is weaker than the one expressed by epistemic können. However, a replacement test shows a different picture. In examples (308)– (310), the epistemic uses of mag can neither be substituted with epistemic kann, nor by epistemic könnte without affecting the interpretation. Interestingly, a replacement would decrease the degree of commitment to the truth, contradicting Fritz (1997: 94) and Zifonun (1997: 1894, 1910). Surprisingly, the most appropriate substitute would be dürfte, which has been identified as an epistemic modal verb that refers to an epistemic probability in Section 2.2.5. The assumption that epistemic mögen carry a modal force that is stronger than a possibility is further confirmed by the occurrence of the sentence adverb gewiss ‘certain’ in example (309), which is usually analysed as an epistemic necessity adverb. It deserves to be mentioned that epistemic mögen behaves like dürfte in other respects as well. Most importantly, it cannot be classified as a well-behaved possibility modal verb, as it fails all of the relevant tests, just as has been shown with dürfte in Section 2.2.5. According to Levinson (2000: 36), epistemic possibility operators induce scalar implicatures. A canonical epistemic possibility operator should thus not cause a contradiction in a configuration in which the possibility operator that takes scope over a proposition is conjoined with the possibility operator that selects the negated proposition such as: (♢p) & (♢¬p). And indeed, the epistemic possibility verbs kann and könnte are acceptable in these patterns, see (311a)– (311b). Similar observations have been made by Papafragou (2006: 1693) and Kotin (2008: 382). In contrast, epistemic mögen yields a contradiction in this environment (311c):

239 DeReKo: BRZ05/OKT.19297 Braunschweiger Zeitung, 21/10/2005.

2.2 Case studies |

(311)

a.

161

Anatol kann den Brief gelesen haben, aber er kann ihn Anatol can the letter read-ppp have-inf but he can it genauso gut auch || nicht gelesen haben. exactly.as well also neg read-ppp have-inf ‘It could be that Anatol has read the letter but it could also be that he has not read it.’

b.

Anatol könnte den Brief gelesen haben, aber er Anatol can-sbjv.pst the letter read-ppp have-inf but he könnte ihn genauso gut auch || nicht gelesen haben. can-sbjv.pst it exactly.as well also neg read-ppp have-inf ‘It could be that Anatol has read the letter but it could also be that he has not read it.’

c.

# Anatol mag den Brief gelesen haben,

aber er mag ihn Anatol may the letter read-ppp have-inf but he may it genauso gut auch || nicht gelesen haben. exactly.as well also neg read-ppp have-inf Intended reading:‘It is possible that Anatol has read the letter but it is also more than possible that he has not read it.’

As it turns out, epistemic mögen does not seem to trigger scalar implicatures at all under the same prosodic pattern. For some reason, epistemic mögen (cf. 312c) is less acceptable in this configuration than the canonical epistemic possibility modal verbs kann (cf. 312a) and könnte (cf. 312b), which behave exactly in the way prediceted by Levinson (2000: 36). (312)

a.

Anatol kann den Brief gelesen haben, aber er muss ihn Anatol can the letter read-ppp have-inf but he must it nicht gelesen haben. neg read-ppp have-inf ‘It could be that Anatol has read the letter but it need not be that he has read it.’

b.

Anatol könnte den Brief gelesen haben, aber er muss Anatol can-sbjv.pst the letter read-ppp have-inf but he must ihn nicht gelesen haben. it neg read-ppp have-inf ‘It could be that Anatol has read the letter but it need not be that he has read it.’

c.

# Anatol mag den Brief gelesen haben,

aber er muss ihn Anatol may the letter read-ppp have-inf but he must it nicht gelesen haben. neg read-ppp have-inf Intended reading: ‘It is possible that Anatol has read the letter but it need not be that he has read it.’

162 | 2 Modal verbs: A class struggle

The contrasts are maintained even if the order of the conjuncts is swapped. (313)

a.

Anatol muss den Brief nicht gelesen haben, aber er kann Anatol must the letter neg read-ppp have-inf but he can ihn gelesen haben. it read-ppp have-inf ‘It could be that Anatol has read the letter but it need not to be that he has read it.’

b.

Anatol muss den Brief nicht gelesen haben, aber er Anatol must the letter neg read-ppp have-inf but he ihn gelesen haben. könnte can-sbjv.pst it read-ppp have-inf ‘It could be that Anatol has read the letter but it need not to be that he has read it.’

c.

# Anatol muss den Brief nicht gelesen haben,

aber er mag Anatol must the letter neg read-ppp have-inf but he may ihn gelesen haben. it read-ppp have-inf

‘It is more than probable that Anatol has read the letter but it need not to be that he has read it.’

As these examples show, epistemic mögen cannot be considered as a well-behaved epistemic possibility modal verb anymore. This insight is reflected in the way epistemic mögen interacts with quantifying NPs. In Section 4.20, it will be shown that typical epistemic possibility modal verbs such as können and könnte can occur in the scope of a universally quantifying subject NP, as is shown in (314a) and (314b). These configurations obtain an interpretation in which the culprit can be identified with any person. Such a reading is not available with mögen (314c): This pattern is restricted to the narrow scope interpretation, in which the culprit is everybody at the same time. (314)

a.

Dieses Schaufenster kann jeder eingeschlagen haben this shop.window can everybody break-ppp have-inf ‘Anybody could have broken this shop window.’

b.

Dieses Schaufenster könnte jeder eingeschlagen this shop.window can-sbjv.pst everybody break-ppp haben have-inf ‘Anybody could have broken this shop window.’

c.

# Dieses Schaufenster mag jeder

this

eingeschlagen haben shop.window may everybody break-ppp have-inf

‘Probably, everybody has broken this shop window.’

2.2 Case studies |

163

This equally holds for the examples like those provided by Huitink (2008) in (315) and the two corpus examples (317) and (319). In either case, a substitution by the epistemic modal verb mögen causes an interpretation that refers to a very unlikely state of affairs. (315)

Mindestens drei Männer können der Vater meines Kindes sein. at.least three men can the father of my child be-inf ‘At least three men could be the father of my child.’

(316)

# Mindestens drei Männer mögen der Vater meines Kindes sein.

at.least

three men

may

the father of

my

child

be-inf Intended reading: ‘Perhaps, at least three men are the father of my child.’

(317)

Da die Schule den ganzen Tag und teilweise auch am späteren as the school the whole day and partially also at later Abend zugänglich ist, kann jeder die Kopien mitgenommen evening accessible is can everyone the copies with.take-ppp haben.²⁴⁰ have-inf ‘As the school is open during the whole day and sometimes until late in the evening, anyone could have taken the copies.’

(318)

# Da die Schule den ganzen Tag und teilweise auch am späteren

as the school the whole day and partially also at later Abend zugänglich ist, mag jeder die Kopien mitgenommen evening accessible is may everyone the copies with.take-ppp haben. have-inf Intended reading: ‘As the school is open during the whole day and sometimes until late in the evening, everyone has perhaps taken the copies.’

(319)

„Diesen Brief könnte jeder geschrieben haben, es geht in this letter could everyone write-ppp have-inf it goes in keine politische Richtung”, so Werner Hosiner-Gradwohl.²⁴¹ no political direction so Werner Hosiner-Gradwohl ‘ “Anyone could have written this letter, it does not indicate any political direction.”, according to Werner Hosiner-Gradwohl.’

240 DeReKo: A98/JUN.37190 St. Galler Tagblatt, 05/06/1998. 241 DeReKo: BVZ07/SEP.03009 Burgenländische Volkszeitung, 26/09/2007.

164 | 2 Modal verbs: A class struggle

(320)

# „Diesen Brief mag jeder

geschrieben haben, es geht in keine this letter may everyone write-ppp have-inf it goes in no politische Richtung”, so Werner Hosiner-Gradwohl. political direction so Werner Hosiner-Gradwohl

‘Intended reading: “Everyone could have written this letter, it does not indicate any political direction.”, according to Werner Hosiner-Gradwohl.’

Finally, examples in which a universally quantifying NP takes scope over epistemic mögen could not be found in the DeReKo corpus.²⁴² What can be concluded from these observations? First of all, they indicate that epistemic mögen cannot be considered as a well-behaved possibility modal verb, similarly to epistemic dürfte. At this point the question arises why epistemic mögen behaves in a different way. On the one hand, one could assume that it is its concessive semantics that renders epistemic mögen unacceptable in all of the contexts that are typical of true possibility modal verbs. But as has been indicated above, epistemic mögen can occasionally occur with a more neutral interpretation that lacks any concessive resonance. In all of the examples above, there is nothing that suppresses that latter type of interpretation. In principle, these instances can always be interpreted with both readings. But even under the neutral epistemic interpretation, these examples do not seem to get any better. Thus, the concessive resonance does not appear to have any influence on the acceptability of epistemic mögen in the environments above. But then, it could turn out that epistemic mögen involves a modal force that is stronger than possibility, in the case of epistemic dürfte. This assumption is supported by the observation that the most appropriate substitute for epistemic mögen is epistemic dürfte in a number of contexts. Furthermore, epistemic mögen can be combined with the sentence adverb gewiss ‘certain’ in (309), which is generally regarded as an adverb of epistemic necessity. Finally, it could be another semantic feature yet to be discovered that decreases the acceptability of mögen in the examples above. Maybe it qualifies the type of premises or evidence the epistemic conclusion is based upon, like in the case of epistemic modal verbs in the subjunctive of the past. Summing up, it has been shown that epistemic mögen is very peculiar in many respects in Contemporary German. It cannot be considered as a prototypical possibility verb anymore, which is somewhat surprising, as Fritz (1997: 9) illustrated that it was the first of the traditional modal verbs that developed an epistemic in-

242 The investigation of the DeReKo corpus was carried out on 22nd November 2012, exploiting the Corpus TAGGED-T based on the queries (jeder /+w3 mag) /s0 (MORPH(VRB pp) sein) and (jeder /+w3 mag) /s0 (MORPH(VRB pp) haben)

2.2 Case studies |

165

terpretation. Moreover, it was one of the most frequent modal verbs until the Early New High German period. At this point, the question arises to what extent these peculiarities already existed in earlier stages of German. Firstly, there is good evidence that the concessive epistemic use of mögen is a rather late development. On the one hand, Fritz (1991: 48) has failed to find examples of it for the 16th century in his corpus. Yet, in a corpus study carried out in the course of the investigation presented here, a plausible candidate from 1537 could be identified, as example (1000) in Section 7.1 illustrates. In this period, the most typical epistemic possibility verb appears to be rather the subjunctive of the past form moechte. This could be an indication that epistemic concessive mag was not yet frequently used in this period. But then, Allard (1975: 69,70) concludes that mögen with concessive resonance is derived from the more neutral use. This corresponds to the scenario for mögen provided by Bech (1951: 23), Fritz (1997: 9) and Diewald (1999: 392), who have demonstrated that mögen used to be the default possibility modal verb until the Early New High German period, when it was gradually replaced by the new possibility modal verb können. In turn, mögen acquired new semantic features, and developed an emotive reading. In order to ensure the semantic integrity of the lexical entry, all of the readings of mögen were affected. This explains why mögen turned into a concessive epistemic modal verb, and why the neutral epistemic reading is about to disappear. This will be furthermore addressed in Chapter 7. However, there is a second issue that cannot be solved that easily. It is not clear whether mögen occurred as a well-behaved epistemic possibility modal verb at all. It still remains to be verified whether there is any evidence that epistemic mögen was acceptable in the environments typical of possibility verb in any earlier stage of German.

2.2.8 möchten As was illustrated in the preceding section, the morphological subjunctive of the past form of mögen has lost its original meaning and has emancipated from the lexical entry of its stem. Accordingly, möchte does not contribute any counterfactual resonance anymore, and it has developed into a volitional verb which resembles wollen in many respects. A similar observation can be found in Lötscher (1991: 338, 354). These facts have motivated a range of authors, such as Öhlschläger (1989: 7), Kiss (1995: 162), Fritz (1997: 103), Diewald (1999: 144), Wurmbrand (2001: 137, 183, 224) and Axel (2001: 40), to assume that möchte is to be seen as an independent lexical item. A similar view is held by Reis (2001: 286), but in a less explicit manner. The case of möchte is reminiscent of English must. As Curme (1931: 410) has

166 | 2 Modal verbs: A class struggle

pointed out, it is a former subjunctive of the past form that has acquired present indicative meaning. Nevertheless, there is no consensus to what extent möchte should be considered as an independent modal verb. This is partly due to the difficulties of finding a consistent definition for the so-called modal verbs in German, as was already shown in Section 2.1. What reasons are there in favour of an analysis as an independent modal verb? First and foremost, the semantic interpretation of möchte cannot be compositionally derived from mögen. Rather, it acquired a meaning that is almost identical to the one of wollen. As Diewald (1999: 147) assumes, the only difference compared to wollen concerns the expectations of the subject referent. In the case of möchte, the subject referent does not insist that the state of affairs expressed by the infinitive will be realised, he leaves the option of giving up on his intention. In a similar vein, Welke (1965: 114) and Vater (2010: 104) observe that möchte sounds more formal and polite than its counterpart wollen. Due to its behaviour, which is almost identical to that of wollen, there is no plausible reason to exclude möchte from the group of modal verbs. If one considers wollen as a genuine modal verb, one has to consider möchte as such as well. Finally, the two volitional verbs share another striking property: both of them are not preterite-presents, but they originated in an old optative or subjunctive form and emancipated from this form. However, there are also arguments against an approach that classifies möchte as a proper and independent modal verb. On the one hand, it is far from obvious to what extent möchte has indeed acquired a complete morphological paradigm, including non-finite and past forms. On the other hand, some authors argue that a proper modal verb has to involve an epistemic interpretation as well. Öhlschläger (1989: 8, 93) excludes möchte explicitly from the class of modal verbs, as he doubts that it can be interpreted in a (reportative) epistemic way. Following the same line of reasoning, Reis (2001: 310) does not appear to consider it as a clear member of that class, either. As recent studies have revealed, however, non-finite uses of möchten are well documented in the spoken language, and partially in written texts. Furthermore, there are some instances of reportative möchte and possibly epistemic möchte as well. Vater (2010: 103) illustrates that möchte had already developed a proper infinitive of its own in the 19th century: The example (321) provided by Vater is taken from a fictional text that reflects the spoken language in Austria. Furthermore, such uses can be easily found on the Internet (322)–(323).

2.2 Case studies |

(321)

167

daß der Muckerl kein’ andere will, wie dich und, selbst, wenn that the Muckerl no other as desires you and even if er eine möchten tat, mich schon af d’allerletzt, das he a want-inf do-sbjv.pst me already on the.last that weißt. . . ²⁴³ know ‘That Muckerl desires no other one than you and even if he wanted one, I would be the last one, you know that.

(322)

Wenn Sie sich bereits entschieden haben, bei uns eintreten zu if you refl already decide have at us join-inf to möchten, sollten Sie folgenden Ablauf beachten:²⁴⁴ want-inf should you following procedure notice-inf ‘In case you have already decided to (want to) join us, you should consider the following procedure.’

(323)

Ohne moralisch werten zu möchten, kann der Monotheismus without morally judge-inf to want-inf can the monotheism nicht mit dem Polytheismus verglichen werden²⁴⁵ neg with the polytheism compare-ppp pass.aux-inf ‘Without having the intention to judge in a moral way, monotheism cannot be compared to polytheism.’

Moreover, the infinitive form möchten can be frequently heard in spoken language by attentive listeners. Similar observations have been made by Eisenberg et al. (2005: 566). It appears, then, that möchte has undergone the same development as wollen and the other modal verbs. It used to be a form of a particular verb that emancipated and became an independent verb on its own. Thus, they all started out with a defective morphological paradigm that lacked, in particular, non-finite forms, and forms for the past. Step by step, they developed the missing forms. Recall that Ebert et al. (1993: 413–414) have demonstrated that none of the traditional six modal verbs had developed a proper past participle until the 13th century. Secondly, as will be shown below in Sections 2.2.8.7 and 2.2.8.8, there are a couple of instances that could be considered as reportative or possibly epistemic usages as well. This indicates that there are good arguments to analyse möchte along the lines of wollen. There are a couple of environments in which möchte

243 Ludwig Anzengruber, Sternsteinhof, p. 42, (1890) 244 http://www.thw-nuernberg.de/mitmachen/ accessed on 1st December 2011. 245 http://www.religionsforum-wogeheichhin.de/t2163f16-Ellinais-Zurueck-zu-GoettervaterZeus.html accessed on 1st December 2011.

168 | 2 Modal verbs: A class struggle

could not be found. It is not obvious why this is the case. This behaviour could be accounted for on the basis of the assumption that möchte is less developed, or less modal, than wollen. Alternatively, it could be explained in terms of frequency. Firstly, volitional möchte occurs far less frequently than volitional wollen. And all of the environments where möchte could not be found turn out to be environments in which wollen is fairly rare as well. The volitional verb möchte behaves in a way almost identical to the one of wollen. All of the exceptional properties that had been observed for wollen appear to hold for möchte as well. There does not appear to be any spectacular behaviour that is particular to möchte. To avoid any unnecessary redundancy, the comments made on this matter in this section will be rather short. To understand the nature of volitional verbs such as möchte and wollen in more detail, the reader is referred to Section 2.2.3, which is devoted wollen. 2.2.8.1 Transitive uses Analogous to wollen, it is expected that möchte used without infinitive should be possible as a transitive verb as well. As it has not developed a proper past participle yet, it is not possible to apply the passive test. Given these observations, it will only be shown here that instances without an infinitive complement exist. It is not possible to directly prove that these occurrences indeed involve transitive patterns; it can just be concluded from the nature of wollen, which is arguably the prototype for volitional möchte. Some authors, such as Raynaud (1977: 5) and Eisenberg (2004: 96), explicitly assume that möchten can occur with NP objects. Instances like (324) or (325) are frequently found in corpora. (324)

94,6 Prozent der Deutschen möchten eine deutlich bessere 96,4 percent the-gen German want a clearly better Kennzeichnung von gentechnisch veränderten Lebensmitteln.²⁴⁶ labelling of genetically manipulated food ‘96.4 percent of the German population want a clearly better labelling of genetically manipulated food.’

(325)

Kinder sind am besten bei Mann und Frau aufgehoben. Eine children are at best at man and woman stored a völlige Gleichstellung möchten wir nicht.²⁴⁷ complete equalisation want we neg ‘Kids should preferably be raised by man and woman. We do not want an entirely equal treatment.’

246 DeReKo: HMP09/APR.01329 Hamburger Morgenpost, 16/04/2009 247 DeReKo: M09/JUL.58113 Mannheimer Morgen, 25/07/2009

2.2 Case studies |

169

In both cases, passivisation is plausible if möchte is replaced with wollen. 2.2.8.2 dass-clauses A whole range of authors, e.g. Welke (1965: 114), Raynaud (1977: 6), Öhlschläger (1989: 70), Reis (2001: 304) and Vater (2010: 105), already observed that möchte can select finite dass-clauses, just as wollen does. Such examples frequently occur in corpora, as is exemplified in (326) and (327): (326)

Es ist egal, was Suri machen möchte, Tom möchte, dass sie es it is equal what Suri make-inf wants Tom wants that she it besser kann als alle anderen.²⁴⁸ better can than all others ‘No matter what Suri wants to do, Tom wants that she does it better than any one else.’

(327)

So möchten 91 Prozent der Frauen, dass ein neuer so want 91 percent the-gen women-gen that a new Verehrer ihnen bereits bei der ersten Verabredung einen Kuss admirer them already at the first date a kiss gibt.²⁴⁹ give ‘Accordingly, 91 percent of the women want that new admirers kiss them already on their first date.’

(328)

Hier bleibt ihr bitte stehen, weil hier möchte ich nicht, here stay you please stand because here-foc want I neg dass ihr alleine rübergeht.²⁵⁰ that you alone over.go ‘Here you have to stop, as I do not want you to cross the street alone here.’

The spoken example (328) provides an interesting case, as it contains an extraction of the locative adverb hier ‘here’, which bears focus across the boundary of the dass-clause. It is remarkable that the extracted VP-adverb hier ‘here’ bears contrastive focus accent. Accordingly, it could be considered as an instance of A′ movement, just as wh-movement. As a consequence, it is reminiscent of extractions out of wh-clauses, which is acceptable at least in some southern varieties of German.

248 DeReKo: HAZ09/APR.01834 Hannoversche Allgemeine, 11/04/2009. 249 DeReKo: M08/FEB.12973 Mannheimer Morgen, 19/02/2008. 250 Mother to her children, Berlin Charlottenburg 2th June. 2012.

170 | 2 Modal verbs: A class struggle

2.2.8.3 Control infinitives with event modification Most notably, möchte occurs with a bare infinitive complement displaying a volitional interpretation. Just as its volitional relative wollen, it is generally held to be a control verb involving a proper subject argument of its own, as demonstrated by Öhlschläger (1989: 119), Kiss (1995: 162), Diewald (1999: 140), Axel (2001: 40), Erb (2001: 78), Reis (2001: 302) and Wurmbrand (2001: 170). Such patterns are very frequently attested in corpora, cf. (329)–(330): (329)

Kinder oder Ehepartner können nichts dagegen unternehmen, children or spouse can nothing against undertake-inf wenn ihr Verwandter seinen Körper plastinieren lassen if their relative his body plastinate-inf letmöchte²⁵¹ wants ‘Children or spouses cannot prevent their relatives from getting plastinated if they want to become plastinated.’

(330)

Ich möchte Sprengmeister werden.²⁵² I want blaster become-inf ‘I want to become a blaster.’

Once again, möchte with bare infinitive is almost synonymous with its counterpart wollen in these examples. 2.2.8.4 Control directionals with event modification As Szumlakowski-Morodo (2006: 330) already pointed out, möchte also takes verbless directional phrase complements. Patterns such as (331) and (332) can be found in the DeReKo corpus: (331)

Die 15-Jährige möchte gern zur Polizei.²⁵³ the 15.year.old wants gladly to.the police ‘The 15 year old would like to join the police.’

(332)

Ein Umzug dürfte notwendig werden, doch kaum jemand a relocation might necessary become-inf but hardly any der verbliebenen Mieter möchte raus aus dem Haus.²⁵⁴ the-gen remaining-gen tenants-gen wants out of the house

251 DeReKo: M09/JAN.02729 Mannheimer Morgen, 13/01/2009. 252 DeReKo: M06/JUL.58784 Mannheimer Morgen, 26/07/2006. 253 DeReKo: HAZ09/JAN.04281 Hannoversche Allgemeine, 27/01/2009.

2.2 Case studies |

171

‘might become possible, but hardly any of the remaining tenants want to quit the house.’

Szumlakowski-Morodo (2006) indicates that verbless directional phrases such as the ones illustrated above do not involve ellipsis of the infinitive. 2.2.8.5 Raising infinitives with event modification As has been been suggested by Wurmbrand (2001: 170), wollen is not the only volitional verb that can be used with weather verbs. Apart from wollen, möchte can also occur in these configurations, which are usually held to imply raising. Similarly to the case of wollen, a couple of different patterns have to be distinguished. First of all, there is the negative polar raising pattern of möchte, as illustrated in examples (333) and (334): (333)

Der Frosch verwandelt sich trotz Mundspray und liebevollem the frog changes refl despite mouth.spray and loving Kuss nicht in den ersehnten Prinzen. Auch bei der Kröte und der kiss neg in the longed.for Prince also with the toad and the Ratte möchte das Vorhaben nicht gelingen.²⁵⁵ rat wants the enterprise neg succeed-inf ‘The frog does not change into the longed for Prince, despite a mouth spray and a tender kiss. The enterprise simply does not to happen to succeed with the toad and the rat either.’

(334)

„[...]So erschöpft und kaputt ist man während einer Etappe, so exhausted and broken is one during a stage wenn der Anstieg einfach nicht aufhören möchte oder der when the climb simply neg end-inf wants or the Schneesturm unbarmherzig tobt”, beschreibt der Hundesportler snowstorm mercilessly blusters describes the dog.sportsman schwache Momente auf dem Schlitten und fährt fort:²⁵⁶ weak moments on the sledge and goes on ‘ “One is that much exhausted and beat-up at some a stage if the climb simply does not happen to end or the snowstorm is mercilessly blustering” the dog sledge pilot describes weak moments on the sledge and continues.’

Just like the typical cases of negative polar raising wollen, the utterance in (334) contains the adverb einfach.

254 DeReKo: BRZ08/OKT.15219 Braunschweiger Zeitung, 30/10/2008. 255 DeReKo: A00/FEB.13085 St. Galler Tagblatt, 21/02/2000. 256 DeReKo:NON09/FEB.10904 Niederösterreichische Nachrichten, 23/02/2009.

172 | 2 Modal verbs: A class struggle

Apart from this usage, möchte occurs as a raising verb in an environment in which it is almost synonymous with the necessity modal verb müssen, just like the respective raising pattern of wollen, which has been discussed by Helbig and Buscha (2001: 121). (335)

Da möchte das Geschriebene erst recht und lückenlos there wants the written yet right and completely entschlüsselt sein.²⁵⁷ decode-ppp be-inf ‘In this case, the writings have to be even more completely decoded.’

(336)

Den Klassenerhalt als Ziel ausgegeben haben die TSG Rheinau the class.sustain as goal defined have the TSG Rheingau und der SC Rot-Weiß Rheinau II, der aber früher gesichert and the SC Red-White Rheinau II the but earlier assure-ppp sein möchte als in der vergangenen Runde.²⁵⁸ be-inf want as in the previous round ‘TSG Rheingau and SC Red-White Rheinau II have defined as their goal to say in the same league, but this has to be assured earlier than in the previous playing time.’

Finally, there are rare instances of möchte that exhibit the old meaning of a possibility modal verb. In the example given below, möchte is more appropriately substituted with könnte, rather than will. (337)

und ehe ich das Licht auslöschte, versuchte ich, ob es mir and before I the light out.put tried I if it me wohl gelingen möchte, die scharfe Spitze ein paar Zoll possibly succeed-inf wants the sharp point a some inch tief in die Brust zu senken. da dieses aber niemals gelingen deep in the chest to sink as this but never succeed-inf wollte, so lachte ich mich zuletzt selbst aus, warf alle want so laughed I refl finally self out threw all hypochondrische Fratzen hinweg, und beschloß zu leben²⁵⁹ hypochondriac grimaces away and decided to live ‘and before I put out the light, I tried if it could succeed to sink the sharp point a couple of inches deep into the chest. But, as this never happened to succeed, I laughed out loudly and threw away all of the hypochondriac grimaces and decided to live.’

257 DeReKo: A99/AUG.59080 St. Galler Tagblatt, 28/08/1999. 258 DeReKo: M09/AUG.64710 Mannheimer Morgen, 19/08/2009. 259 DeReKo: GOE/AGD.00000 Goethe: Dichtung und Wahrheit.

2.2 Case studies |

173

2.2.8.6 Raising directionals with event modification In opposition to wollen, the raising pattern of möchte could not be found in the DeReKo corpus. However, it is not evident whether this is due to a lower degree of grammaticalisation of möchte, or rather to the fact that möchte is by far less frequent than wollen. 2.2.8.7 Control infinitives with clause modification Whereas Öhlschläger (1989: 93) assumes that möchte cannot be interpreted in a reportative way, Vater (2010: 107) provides the interesting example (338), where möchte appears to exhibit a reportative interpretation. An analogous example could also be found in the DeReKO corpus (cf. 339) and in another text (cf. 340). (338)

Niemand aus der Nachbarschaft will etwas bemerkt nobody from the neighbourhood wants something notice-ppp haben, keiner möchte auch nur Verdacht geschöpft haben.²⁶⁰ have-inf nobody wants also only suspicion scoop-ppp have-inf ‘All of the neighbours claim that they had not noticed anything, all of them claim that they did not have any suspicion.’

(339)

Keine Bedenken gegen das neue Einkaufszentrum zu haben no objections against the new shopping.mall to have-inf möchte Gerda Stecker jedoch nicht gesagt haben.²⁶¹ wants Gerda Stecker yet neg say-ppp have-inf ‘Gerda Stecker claims that she has not said, she has any objections against the new shopping mall.’

(340)

Wie die Milizen im einzelnen zusammengesetzt waren, ist if the militia in.the only composed were is umstritten, weil keine der bekannten Gruppierungen contested because none the-gen known-gen groups-gen dabeigewesen sein möchte.²⁶² be.present-ppp be-inf wants ‘It is contested how the militias were composed because all of the known groups claim to not have been involved.’

All of the instances of möchte in the examples above embed a predication that refers to a past event. Accordingly, they behave like canonical reportative modal verbs. Even if they do so, it deserves closer attention that all of them occur with

260 Kölner Stadtanzeiger, 30/08/2006, as quoted in Vater (2010: 107). 261 DeReKo: V99/NOV.52643 Vorarlberger Nachrichten, 04/11/1999. 262 Noam Chomsky Offene Wunde Nahost, translated by Michael Haupt, Europa Verlag, p. 167.

174 | 2 Modal verbs: A class struggle

negation. As illustrated in Section 2.2.3 and 2.2.6, it is not easy to determine the scope of negation with a volitional modal verb. Likewise, the negation behaves in a somewhat peculiar way in the examples given above that involve a negative quantifier in subject position (cf. 338 and 340). Ehrich (2001: 167) assumes that reportative modal verbs in German always bear scope over a negative operator, as such configurations provide the more plausible interpretation for reportative modals: x claims that not (p) rather than x does not claim that p. In a similar manner, Öhlschläger (1989: 93) concludes that the narrow scope interpretation for reportative modal verbs is by far the more natural one. However, he does not want to exclude configurations in which the negation takes scope over the reportative modal verb. Examples (338) and (340), in which möchte involves a negative subject quantifier, are essential challenges for both approaches. Being a control verb, möchte requires a subject referent. Accordingly, the subject NP cannot have entirely raised from the infinitival clause. Hence, there are three options for how the interpretation comes about. Firstly, one could assume that the negative subject NP has not raised at all. Accordingly, the negation should be interpreted in the subject position, and the corresponding gloss would be as shown in (341). This does not seem to be the case. Alternatively, one could argue that only the negative particle has been raised into the subject position, where it has attached to the subject NP. Given that the decomposition of the negative quantifier niemand ‘nobody’ and keiner ‘no’ results in the combination of operators ¬∃ rather than ∀¬, the expected translation should be as in (342). Once again, this is not the case. The most appropriate gloss is the last one (cf. 343). (341)

# Nobody claims to have had a suspicion.

(342)

# Somebody claims not to have had any suspicion.

(343)

Everybody claims not to have had any suspicion.

Finally, one could argue that the reportative uses of wollen and möchte cover a broader range of interpretation than is usually assumed. Besides its reference to a claim, it could possibly denote a confession as well. This would be appropriate in the examples given above. Yet, it remains to be explained why this shift occurs only if a negation is present. Apart from that, it appears to be plausible that the strange interaction is a result of an implicature, just as in the case of sollen, as was pointed out in Section 2.2.6.2. Unless möchte is not found in environments without negation, we should not jump to the conclusion that it can be considered as a reportative modal verb without any restriction.

2.2 Case studies |

175

2.2.8.8 Raising infinitives with clause modification Finally, there are a couple of instances which appear to be interpreted in an epistemic way. According to Vater (2010: 107), epistemic instances of möchte can be observed in Saxonian dialects, cf. (344). (344)

a.

A: Kommt Paul morgen? comes Paul tomorrow

b.

B: Das möchte schon sein. That might par be-inf A: ‘Does Paul come tomorrow?’ B: ‘It could be.’

In example (344) provided by Vater (2010: 107), it is not obvious to what extent the instance of möchte is indeed derived from its volitional use. If this is the case, it should be possible to substitute it by its volitional relative wollen without changing the communicative effect. Alternatively, this occurrence of möchte reflects the old use as a possibility verb. In this case, its appropriate substitute would be könnte. Occasionally, instances of epistemic möchte can be found where this verb is more adequately replaced with könnte, as in the pattern in (345) given by Zifonun (1997: 1270), and in example from the DeReKo corpus (346). (345)

Nicht so sehr von dem Gedanken, daß Eduard unter neg so much by the thought that Eduard under Umständen für ihre Ehre sein junges Leben möchte in die circumstances for her honour his young live wants in the Schanze zu schlagen haben,– die romantische Vorstellung entrenchment to beat-inf have-inf the romantic image obgleich sie darüber geweint hatte, ließ ihr Herz eher höher even.if she there cried had let her heard rather higher schlagen.²⁶³ beat-inf ‘It was not so much this thought that Eduard possibly would risk his life to defend her honour, it was rather the romantic image that made her heart beat faster – even if she had cried about it.’

263 Thomas Mann, Erzählungen volume 8, as cited in Zifonun (1997: 1270).

176 | 2 Modal verbs: A class struggle

(346)

Erbrochenes, Schweiß, Kot und Sperma ziehen sich durch vomit sweat faeces and sperm draw refl through das Werk, als wär’s ein neuer Megaseller von Charlotte the act if is-sbjv.pst a new bestseller by Charlotte Roche. Ist es aber nicht. Gottlob, möchte man meinen.²⁶⁴ Roche is it but neg thank.god could one opine-inf ‘The book is full of vomit, sweat, faeces and sperm, as if it was a new bestseller by Charlotte Roche. But it is not. “Thank God”, the readers could opine.’

Epistemic möchte does not occur very frequently with a possibility interpretation anymore in German. It is mostly restricted to idiomatic collocations such as möchte meinen ‘could believe’ (cf. 346). The context of this example is rather opaque. A similar occurrence of möchte is discussed by Welke (1965: 116).

2.2.9 brauchen In contrast to the items discussed so far, brauchen ‘need’ has a very different origin with respect to its morphological paradigm. It is neither a preterite-present, nor did it develop from a former optative form, as in the case of wollen. Nevertheless, it exhibits a whole range of properties which are typical of the six traditional modal verbs. However, its precise status is contested. On the one hand, grammarians such as Engel (1996: 463), Weinrich (1993: 300) and Szumlakowski (2010: 79) regard brauchen as a fully developed modal verb, while others, such as Brinkmann (1962: 363) and Öhlschläger (1989: 8), explicitly exclude brauchen from the class of modal verbs. Less clearly, the latest edition of the popular Duden grammar edited by Eisenberg et al. (2005: 562) appears to be inclined to consider brauchen as a modal verb: “brauchen ist seiner Verwendung nach den Modalverben zuzurechen” (‘Regarding its uses, brauchen has to be considered as modal verb’). What arguments are there in favour of the view that brauchen should be considered as a modal verb? There are at least six reasons. Firstly, it shares an essential morphological peculiarity with the traditional six modal verbs: (i) In perfect tense, the past participle selected by the tense auxiliary haben ‘have’ is formally identical to the infinitive. This phenomenon is referred to as infinitivus pro participio; the (IPP)-effect and was discussed at length in Section 2.1.1.2. (ii) Kolb (1964: 74) has shown that brauchen unexpectedly acquired the irregular subjunctive of the past form bräuchte ‘need-sbjv.pst’, which carries an Umlaut. Originally, brauchen is a regular verb, which do not carry an Umlaut in their past subjunctive forms. Moreover, there are syntactic similarities. Folsom (1968: 322–323) and 264 DeReKo: HAZ09/JAN.02665 Hannoversche Allgemeine, 19/01/2009.

2.2 Case studies |

177

Szumlakowski (2010) have pointed out that brauchen can sometimes occur with (iii) a bare infinitive complement and (iv) verbless directional phrases, just as the traditional six modal verbs do. Apart from that, (v) brauchen is restricted to negative environments. In this respect it behaves like a whole range of other modal auxiliaries, such as raising wollen (cf. Section 2.2.3.5) and volitional mögen (cf. Section 2.2.7). Furthermore, Bech (1951: 14), Kolb (1964: 73) and Lenz (1996: 399) have shown that brauchen is just as negative polar as its predecessor dürfen (cf. Section 2.2.4). Finally, there are also semantic reasons that speak for an analysis of brauchen as a modal verb. As Kolb (1964: 74), Lenz (1996: 402) and Askedal (1997a: 61) illustrate, (vi) nicht brauchen ‘not need’ denotes a negated circumstantial necessity, just like nicht müssen. In addition, (vii) Takahaši (1984), Askedal (1997a: 62) and Reis (2005a: 112) point out that there are instances of brauchen that appear to involve an epistemic interpretation. By contrast, there are a couple of considerations that led other authors to the conclusion that brauchen should not be fully considered as a modal verb in German. Some authors, such as Brinkmann (1962: 363) and Maitz and Tronka (2009: 189), argue that brauchen differs crucially from the traditional modal verbs in various respects: (i) It selects zu-infinitive complements rather than bare infinitive complements, and (ii) it is restricted to negative polarity environments. However, as has been illustrated above, negative polarity is a property that occurs with three out of six members of the traditional modal verb class: volitional mögen, the raising use of wollen and dürfen up to the 16th century. Furthermore, Paul (1897: 79) demonstrates that brauchen has replaced dürfen in its original meaning. This indicates that negative polarity is rather a property typical of modal verbs. Apart from that, other authors, such as Folsom (1968: 328) and Öhlschläger (1989: 8), challenge the existence of an epistemic interpretation. Yet, these authors have not taken the corpus examples provided by Takahaši (1984) into account. As will be shown below, it is far from evident whether or not brauchen has a well established epistemic interpretation. As it turns out, most of the remaining Germanic languages have an equivalent verb that reflects the situation in German almost exactly. Mortelmans, Boye and Auwera (2009: 17) illustrate that in Dutch hoeven ‘need’ usually selects te infinitive complements. Sometimes, it can be found with bare infinitive complements. Just like its German counterpart brauchen it is restricted to negative polarity environments. As Scaffidi-Abbate (1973: 14) points out, Afrikaans bruik exhibits almost the same property. Yet, he considers German influence fairly unlikely. As Kolb (1964: 76) and Scaffidi-Abbate (1973: 30) indicate, the verb need in English is a negative polarity item as soon as it selects infinitive complements. Similarly to brauchen, it was originally restricted to to-infinitive complements, rather than bare infinitive complements. In the course of history, bare infinitives became

178 | 2 Modal verbs: A class struggle

possible as well. These observations can be extended to Northern Germanic languages as well. Eide (2005: 77) has demonstrated that behøve ‘need’ and trenge ‘need’ in Norwegian can be used to express a necessity. In this use, these verbs are negative polarity items and alternate between å-infinitive complements and bare infinitive complements. In a similar fashion, Mortelmans, Boye and Auwera (2009: 42) show that Danish behøve ‘need’ originally only selected at infinitive complements. At some later period, bare infinitive complements became possible as well. As Paulina Tovo (pers. commun.) has pointed out, it is restricted to negative environments in this use. As it turns out, all of these languages, which are genetically related to each other, have developed analogous patterns. Yet, the most astonishing fact is that almost each language has chosen a verb with a different etymological origin to adopt the role of brauchen. As Scaffidi-Abbate (1973: 20) remarks, this behaviour is parallel to Latin uti, which originally used to be restricted to negative environments. This discussion about the precise status of brauchen with respect to the modal/auxiliary verbs has existed since the 19th century. As was illustrated in Section 2.1.1.2, Grimm (1837: 168, 949) is reluctant to consider brauchen as an auxiliary verb. According to him, verbs that select zu-infinitive complements are generally incompatible with the IPP-effect and must not be used in this way. In a note on page 949, he discusses an example of brauchen with zu-infinitive complement that exhibits the IPP-effect. By contrast, Becker (1836: 177 §91, 1842: 220 §91, 224 §93) explicitly classifies brauchen with an infinitive complement as a Huelfsverb des Modus ‘auxiliary of mood’. He only considers cases in which brauchen selects zu-infinitive complements. And the fact that this configuration differs from more canonical modal verbs such as müssen does not appear to bother him. Decades later, Sanders (1908: 101), who is another normative grammarian, considers brauchen with the IPP-effect as fully grammatical and observes that the infinitive particle zu can be dropped under certain conditions even in formal language. In the remainder of this section, the multitude of the different uses of brauchen will be discussed. Above of all, brauchen is used as a transitive verb that selects an accusative NP. In this use, it is not restricted to negative polarity environments. Moreover, it is occasionally found with finite dass-clauses. Furthermore, it can be used in an impersonal pattern with a non-referential subject NP. Most notably, it is used with a raising infinitive complement that exhibits a circumstantial interpretation. Next to a practical necessity reading, it can be used as a quantificational modal verb as well, just like können and nicht müssen. In the most canonical cases, it involves a subject-to-subject raising configuration, in some varieties a subject-to-object (AcI/ECM) pattern seems to be possible as well. Moreover, it can be combined with verbless directional phrases. Finally, it is very rarely recor-

2.2 Case studies |

179

ded with an epistemic interpretation. As it turns out, brauchen behaves, in many respects, like the most prominent members of the traditional six modal verbs. 2.2.9.1 Transitive uses As is well known, brauchen is a transitive verb. Paul (1897: 79), Kolb (1964: 65), ¯ ¯ Scaffidi-Abbate (1973: 11, 19) illustrate that bruhhan/br uhhen in Old High German originally referred to the enjoyment or the consumption of something just as the English verbs ‘to enjoy/consume’ can be. In this period, it selected a genitive NP. ¯ ¯ Gradually, bruhhan/br uhhen developed a new meaning expressing the usage of an object. At the same time, it can be found with an accusative NP for the first time. Paul (1897: 79) and Scaffidi-Abbate (1973: 19) further demonstrate that brauchen acquired a new meaning in the 17th century. In texts from that period, it occasionally expresses a need or requirement. It merits closer attention that the new meaning of transitive brauchen was initially restricted to negative contexts, as Paul (1897: 79), Scaffidi-Abbate (1973: 19) and Kluge (2011: 147) have pointed out. In the course of time, transitive brauchen became acceptable in non-negative environments as well. With the latter meaning, it can frequently be found. Interestingly, the transitive use with an accusative NP drastically outweighs the uses with an infinitive. The examples in (347)–(351) reflect the range of different uses and morphological peculiarities. (347)

1,5 Kilogramm brauchte der arbeitslose Drogenabhängige für 1.5 kilo.gramme needed the unemployed drug.addict for den Eigenbedarf.²⁶⁵ the personal.need ‘The unemployed drug addict needed 1.5 kilogramme for personal need.’

(348)

Dass der Mann Geld gebraucht hat, wissen wir, sonst that the man money need-ppp(ge) has know we otherwise hätte er den letzten Banküberfall nicht begangen.²⁶⁶ had he the last bank.robbery neg committed ‘We know that the man needed money. Had the man not needed money he would not have committed the last bank robbery otherwise.’

265 DeReKo: E99/OKT.27727 Zürcher Tagesanzeiger, 22/10/1999. 266 DeReKo: N00/JUN.25283 Salzburger Nachrichten, 03/06/2000.

180 | 2 Modal verbs: A class struggle

(349)

„Aber wollte Mannheim von der Industrialisierung nicht but want-sbjv.pst Mannheim from the industrialisation neg abgehängt werden und zurückfallen, brauchte es outdistance-ppp pass.aux-inf and fall.back-inf need-sbjv.pst it neue Flächen”, verdeutlichte Probst.²⁶⁷ new surface clarified Propst ‘ “If Mannheim did not want to be outdistanced from the Industrialisation and fall back, it would need new surfaces” Probst clarified.’

(350)

Ein Personenzug bräuchte für die rund elf a passenger.train need-sbjv.pst for the about eleven Bahnkilometer nach Innsbruck hingegen nur elf Minuten.²⁶⁸ train.kilometres to Innsbruck however only eleven minutes ‘However, it takes a passenger train only eleven minutes to get the eleven kilometres to Innsbruck.’

(351)

Zwölf Jahre braucht er, um die Sonne einmal zu twelve years need he in.order the sun once to umrunden.²⁶⁹ orbit-inf ‘It takes it twelve years to orbit the sun once.’

Crucially, the past participle of transitive brauchen is realised as a canonical geparticiple in example (348). In this environment, it cannot be substituted with the IPP brauchen. As indicated above, there are two alternatives of realising the subjunctive past form. The regular form illustrated in (349) is identical to the indicative past tense form brauchte. As the conditional configuration given in (349) requires the subjunctive of the past, it becomes obvious that brauchte is interpreted as a subjunctive of the past, rather than the indicative of the past. Apart from that, there is also the irregular form bräuchte, which differs from the weak form in that it carries an Umlaut, as is demonstrated in example (350). This illustrates that the two morphological peculiarities of brauchen behave in a different way: While the IPP is restricted to the uses of brauchen with infinitive complements, the strong form of the past subjunctive is also available for the transitive uses of brauchen. Finally, there is a related use of brauchen in which it selects an NP, a PP or a AP, and where expresses a time interval or other measures that are necessary

267 DeReKo: M04/JUL.45250 Mannheimer Morgen, 07/07/2004. 268 DeReKo: I97/MAI.16915 Tiroler Tageszeitung, 03/05/1997. 269 DeReKo: HAZ09/AUG.04419 Hannoversche Allgemeine, 28/08/2009.

2.2 Case studies |

181

to achieve a certain goal.²⁷⁰ While the measure can be realised as an accusative NP, the goal can be either realised as a PP (cf. nach Innsbruck ‘to Innsbruck’ in example 350) or as an adverbial um zu-infinitive (cf. 351). But the measure phrase can also be realised as an adverb such as lange ‘long’. Furthermore, Stefan Müller (pers. commun.) has observed that the NP which expresses the measure does not agree with the passive auxiliary if brauchen is passivised. Rather, the passive auxiliary always exhibits the default morphology, which is 3rd pers. singular. (352)

[. . . ] weil dafür mehrere Tage gebraucht because therefore several days-acc need-ppp wurde/*wurden. pass.aux-sg/ pass.aux-pl ‘[. . . ] because it took several days for that.’

These facts indicate that this use of brauchen has to be considered as an independent syntactic pattern. 2.2.9.2 Impersonal brauchen Brauchen can be used as an impersonal verb that lacks a referential subject argument. This use was already mentioned by Adelung (1793: 1162), Grimm and Grimm (1860: 318) and Paul (1897: 79). Their observation is mainly based on evidence from writers of the 18th century. Brauchen is only found with the new interpretation, in which it refers to a need. In contrast to the transitive use, it selects the non-referential pronoun es ‘it’ as a subject argument and an accusative NP. Accordingly, the need is not attributed to the subject referent. Rather, the bearer of the need remains unspecified. In all of the examples (353)–(355), the subject NP es is not a referential pronoun: first, it does not identify the bearer of the need and second, it does not refer to a neuter noun that has been mentioned in prior discourse. In example (353), there is no salient antecedent NP specified for neuter in the preceding context. The referent to which the need is attributed is anybody who has to wait for a longer period. Accordingly, the impersonal variant of brauchen was chosen, in order to leave the bearer of the need as unspecified and generic as possible. (353)

Werner Ignaz Jans ist ein ungeduldiger Mensch. Und Geduld Werner Ignaz Jans is a impatient man and patience bräuchte es eigentlich, um ein guter Warter zu sein.²⁷¹ need-sbjv.pst it actually in.order a good waiter to be-inf

270 That this use should be considered as an independent use rather than a variant of the transitive use was pointed out to me by Stefan Müller (pers. commun.).

182 | 2 Modal verbs: A class struggle

‘Werner Ignaz Jans is an impatient man. And it is patience that would be actually necessary in order to wait a long time.’

(354)

Es sieht nicht gut aus für den Frieden in Nahost. it sees neg good out for the piece in middle.east Wahrscheinlich braucht es ein viertes Wunder, damit er eine probably need it a fourth wonder in.order.to he a neue Chance erhält.²⁷² new chance gets ‘There is not much hope for the peace in the Middle East. Probably, a fourth wonder is necessary in order to give it a new chance.’

(355)

Was braucht es außer dem, als daß sie selbst dich liebt?²⁷³ what needs it apart that as that she self you loves? ‘What else is necessary as the circumstance that she loves you?’

Interestingly, the impersonal variant of brauchen found in DeReKo primarily occurs in Swiss newspapers. This high frequency of this pattern in Swiss German could be due to language contact: There is a similar pattern in French that corresponds almost one-to-one to its German counterpart: il faut ‘it necessitates’. Yet, the use of impersonal brauchen is not restricted to texts from Switzerland. Occasionally, it can be found in newspapers from other regions as well, as an occurrence taken from a news paper from Lower Saxony indicates (cf. 354). Moreover, this pattern can already be found in the early 18th century, in a poem written by Gottsched (cf. 355). According to Grimm and Grimm (1860: 318), Kluempers (1997: 87) provides an example exhibiting the impersonal use. It is attributed Johann Balthasar Schupp, who lived from 1610–1661.²⁷⁴ (356)

derowegen braucht es mühe (kostet es m.) dasz du wahre therefore need it effort (cost it effort) that you true freund erwehlest.²⁷⁵ friend chose ‘Therefore, effort is required to chose true friends’

271 DeReKo: A01/NOV.42609 St. Galler Tagblatt, 13/11/2001. 272 DeReKo: BRZ06/JAN.09059 Braunschweiger Zeitung, 19/01/2006. 273 Johann Joachim Schwaben Herrn Johann Christoph Gottscheds Gedichte XVI. Schreiben an Hrn. D. Gottfr. Thomas Ludewig, bey seiner Verheirathung 1732. Leipzig, Breitkopf (1751), p. 408. 274 Kluempers (1997: 87) erroneously ascribes the example to Hans Jakob Christoffel von Grimmelshausen, the author of Der Abentheuerliche Simplicissimus Teutsch (1668) – but this is a result of a misinterpretation of Grimm’s system of labelling the sources. 275 Schuppius, 756 (around 1650), as quoted in Grimm and Grimm (1860: 318).

2.2 Case studies |

183

This is in accordance with the hypothesis that this pattern emerged due to French influence. As Bloch and Wartburg (1986: 252) demonstrate, the verb faillir ‘need’ developed the impersonal pattern il faut in the 14th century. At that time it spread consecutively and became a dominant item to express a need. In turn, the existence of impersonal brauchen possibly had an impact on the development of brauchen with an infinitive. As will be shown below, brauchen with an infinitive involves a raising pattern which is characterised by the lack of the subject argument. It could turn out that the impersonal use of brauchen was a prerequisite for the development of the pattern with a raising infinitive. Yet, it is fairly likely that brauchen was not directly influenced by the French verb faillir. Scaffidi-Abbate (1973: 27) has pointed out that its predecessor dürfen in its old interpretation (‘need’) had already developed an analogous impersonal pattern in the early 16th century. In this pattern, the object of dürfen is realised as a genitive NP. One of the crucial examples provided by the DWB is given in (199) on p. 115 in Section 2.2.4. Only in a subsequent period was the impersonal use of dürfen replaced with its prefixed counterpart bedürfen. Finally, after brauchen adopted the old patterns of dürfen, it acquired an impersonal use as well. It is quite likely that it initially selected a genitive NP in that configuration, which were replaced with an accusative NP in the course of history. 2.2.9.3 dass-clauses Brauchen can occasionally select finite dass-clauses. It appears that this use is ˜ restricted to certain registers, as it could not be found in the DeReKo corpus.²⁷⁶By contrast, it can be frequently found on the web, as is exemplified in (357). (357)

Ich brauche nicht, dass meine Fingerabdrücke irgendwo I need neg that my fingerprints somewhere abgespeichert sind, solange das nicht von der Polizei store-ppp are, as.long.as that neg by the police zwangsweise aufgenommen wurde.²⁷⁷ compulsorily record-ppp pass.aux.pst ‘I do not need that my fingerprints are recorded unless this has been requested by the police.’

It merits closer attention that utterances such as (357) are significantly more acceptable if they involve a negative operator. It seems, then, that brauchen with 276 The research was carried out on 20th June 2012 and it was based an the query &brauchen ‘‘nicht’’ dass. 277 http://www.computerbase.de/forum/archive/index.php/t-842737.html, accessed on 23th May 2012.

184 | 2 Modal verbs: A class struggle

finite dass-clause is a negative polarity item, just like its cognate which takes infinitive complements. This illustrates that there is a strong link between the use that embeds non-finite clauses and the one that embeds finite clauses. This is reminiscent of the volitional uses of mögen: both uses that embed a finite dass-clause or a control infinitive are negative polar. Interestingly, dürfen, in its old interpretation (‘need’), is documented in such a configuration as well, as Scaffidi-Abbate (1973: 27) has demonstrated. 2.2.9.4 Raising infinitives with event modification The most thoroughly discussed use of brauchen is the one with an infinitival complement. A whole range of authors have pointed out that it exhibits numerous parallels with the six traditional modal verbs in morphological, syntactic and semantic respect. Becker (1836: 177 §91) is the first one to explicitly consider brauchen with zu-infinitive as a ‘mood auxiliary’. In the first edition of his grammar, Grimm (1837: 949) observed that brauchen with zu-infinitive can exhibit the IPP-effect. As one of the first scholars, Paul (1897: 79) associates brauchen with an infinitive complement with the class of preterite presents/traditional modal verbs. As he states, brauchen has replaced dürfen in its original use, which referred to a need or requirement.²⁷⁸ In the same period, Sanders (1908: 101) advised using brauchen with a bare infinitive complement. In its first small description, Bech (1955: 210–212) notices that brauchen is semantically equivalent to müssen and furthermore provides some corpus examples from standard written language. The two studies by Kolb (1964) and Folsom (1968) have pointed out even more parallels with the traditional modal verbs in syntactic and semantic respects. More recent studies, such as Takahaši (1984), Askedal (1997a) and Reis (2005a), deal with the extent to which brauchen allows for an epistemic interpretation. They will be taken into consideration in the remainder of this section. The uses of brauchen with an infinitive turn out to be fairly heterogeneous. This is due to the process of grammaticalisation by which this verb is affected. In order to capture the uses of brauchen in the most efficient way, it becomes necessary to consider its development as well. Thus, the following section will take a diachronic perspective at places. Circumstantial brauchen with a raising infinitive complement occurs in the DeReKo corpus far less frequently than its transitive counterpart. A similar observation was made by Pfeffer (1973: 90) for both written and spoken language. As pointed out by Kolb (1964: 64), Folsom (1968: 328), Scaffidi-Abbate (1973: 31), Askedal (1997a: 53) and Reis (2005a: 104), normative grammarians do not toler-

278 Original quotation: In dieser Verwendung ist brauchen an die Stelle von dürfen getreten.

2.2 Case studies |

185

ate the use of brauchen with a bare infinitive complement, they only consider the use with zu-infinitive complement acceptable. A similar view is held by Jäger (1968: 332). In the course of the 20th century, brauchen with a bare infinitive complement was stigmatised as inferior German. Accordingly, brauchen is predominantly used with zu-infinitives in written language. Likewise, Pfeffer (1973: 91) could not find brauchen with bare infinitive complements in his corpus of written language. (358)

Glücklicherweise brauchte niemand evakuiert zu werden, Luckily needed nobody evacuate-ppp to pass.aux-inf denn in die Zimmer der Bewohner war der Rauch nicht because in the room the-gen tenants-gen was the smoke neg gelangt.²⁷⁹ get-ppp ‘Luckily, nobody needed to be evacuated as the smoke did not get into the rooms of the tenants.’

(359)

Der kostbare Schatz bräuchte nur gehoben zu werden.²⁸⁰ the precious treasure need-sbjv.pst only heave-ppp to pass.aux ‘The precious treasure would just need to be heaved.’

(360)

Man braucht nicht weit zu sehn, viel Jammer und Gefahr.²⁸¹ one need neg far to see-inf much misery and danger ‘One does not need to look far for much misery and danger.’

(361)

Man siehet aber auch dabey, daß die Schmoschen nicht enge zu one sees but also there that the loop neg tight to seyn brauchen.²⁸² be-inf need ‘One can see here, that the loops do not need to be tight.’

(362)

Ein Frauenzimmer braucht nicht gelehrt zu seyn.²⁸³ a woman.room need neg adept to be-inf ‘A woman does not need to be adept.’

(363)

Wir brauchen ja nicht zu lieben²⁸⁴ we need part neg to love-inf

279 DeReKo: RHZ09/DEZ.06729 Rhein-Zeitung, 07/12/2009. 280 DeReKo: P97/APR.13472 Die Presse, 08/04/1997. 281 Johann Christian Günther Ode an Herrn Marckard von Riedenhausen Ivris Vtrivsqve Cvltor (1720) in Johann Christian Günther Werke edited by Reiner Bölhoff, Frankfurt: Deutscher Klassiker Verlag, p.620. 282 Kurtzer doch gruendlicher Begriff der Edlen Jaegerey. Nordhausen: Groß (1730), p. 498. 283 Christian Fürchtegott Gellert, Die Betschwester, II. Aufzug 3 Aufzug (1745).

186 | 2 Modal verbs: A class struggle

‘We do not need to love.’

(364)

Wir brauchen nur verstellt zu weinen: So thun sie ihre we need only feignedly to cry-inf so do they their Schuldigkeit.²⁸⁵ guiltiness ‘We only need to cry feignedly and they will do their part.’

(365)

Man braucht nur in dem großen Buche der Welt lesen zu one need only in the big book the-gen world read-inf to können, und man wird auf den meisten Seiten desselben so viel can and one will at the first pages of.it so much Anschweifung finden, daß man fast nur ein glücklicher inspiration find-inf that one almost only a happy Abschreiber zu seyn braucht,²⁸⁶ copyist to be-inf needs ‘One only needs to be able to read in the big book of the world and one will find so much inspiration in these pages that one almost only needs to be a lucky copyist.’

(366)

ATTINGHAUSEN: [. . . ] Unter der Erde schon liegt meine Zeit / under the earth already lies my time Wohl dem, der mit der neuen nicht mehr / braucht zu leben!²⁸⁷ well him who with the new neg more needs to live-inf ‘ATTINGHAUSEN: [. . . ] My age has long been laid beneath the sod / Happy the man, who may not live to see / What shall be done by those that follow me!.’

As examples (360) and (361) indicate, brauchen with an infinitive can already be observed from the early 18th century onwards. Further early examples are provided by Adelung (1793: 1162) and the DWB in Grimm and Grimm (1860: 318), see (362)–(364). Moreover, brauchen occurs fairly often in works written by Lessing (1729–1781). Moreover, there are early instances of prefixed gebrauchen ‘use’ + zu infinitive with the same interpretation like brauchen, as is illustrated in (367). This is interesting, as gebrauchen has retained its original meaning ‘to use’.

284 Christian Fürchtegott Gellert, Die Zärtlichen Schwestern, I. Aufzug. 6. Auftritt (1747). 285 Christian Fürchtegott Gellert, Das Orakel 1. Aufzug 4. Auftritt, (1747). 286 Johann Andreas Cramer, Der Nordische Aufseher Kopenhagen und Leipzig: Johann Benjamin Ackermann, (1758) sechstes Stueck, p. 57. 287 Friedrich von Schiller, Wilhelm Tell II.1 (1805).

2.2 Case studies |

(367)

187

An den beyden Seiten g g ist dieses Behaeltnis leer geblieben, at the both sides g g is this container empty stay damit man neben dem Lohbeete etwas hoehere in.order.to one next the greenhouse something high-comp Baeume stellen koenne; denn da ein solches Caldarium nicht hoch trees put can since as a such caldarium neg high zu seyn gebraucht, so ist es auch oben durch eine besondere to be-inf needs so is it also up by a particular bretterne, hinten abfallende und auf der Mauer c ruhende Decke wooden back gradient and at the wall c resting ceiling abgeschoren worden.²⁸⁸ shave-ppp pass.aux ‘This container remains empty at both sides g g, so we can set the trees next to the underground greenhouse higher. Since such a caldarium does not need to be high, it has been covered by a particular gradient wooden ceiling that rests on the wall c.’

Sanders (1908: 101) and Wustmann (1908: 354) recommend refraining from using (transitive) gebrauchen in order to refer to a need for something. According to them, the correct use of gebrauchen expresses, rather, a usage of something. A small corpus study investigating ancient Greek and Latin epic poetry translated by Johann Heinrich Voß, comprising Homer’s Odyssee and Illias, Ovid’s Metamorphoses and Vergil’s Äneide, indicates that all of the occurrences of brauchen with an infinitival complement found in these texts select a zu-infinitive, and they are negative polarity items. In sum, four instances of brauchen could be found. Surprisingly, it did not occur in the Äneide. (368) Jetzo, Telemachos, brauchst du dich keineswegs zu scheuen!²⁸⁹

now Telemachos need

you refl by.no.means to dread-inf

‘’Now, Telemachos you do not need by no means to dread.

(369) Aber das wißt ihr selber;

was brauch’ ich die Mutter zu but that know you yourselve what need I the mother to loben?²⁹⁰ praise-inf

‘But you know it yourself; what urges me to praise my mother?’

288 Freiherr von Otto Münchhausen, Monathliche Beschaeftigungen fuer einen Baum= und Plantagen=Gaertner, Hannover (1771), p. 202. 289 Odyssee, III, 14 translated by Johann Heinrich Voß, (1781). 290 Odyssee, XXI, 110 translated by Johann Heinrich Voß, 1781.

188 | 2 Modal verbs: A class struggle

(370) Alter

sie weiß es schon; du brauchst dich nicht zu old.man she knows it already you need refl neg to bemühen.²⁹¹ seek-inf

‘Old man, she already knows it, you do not need to make an effort.’

(371) Deine Tapferkeit kenn’ ich; was brauchtest du dieses zu sagen?²⁹²

your courage

know I

what needed

you this

to say-inf

‘I know your courage, what urged you to speak these words?’

(372) Woll’ auch diesen verzeihn! – Für uns nicht brauchst du zu

want also them forgive-inf beten!²⁹³ pray-inf

for us neg need

you to

‘Forgive them, too – You do not need to pray for us.’

(373) O ihr Himmlischen alle gemeinsam / Rief

er aus, unwissend, oh you divine all together shouted he out unaware nicht alle sie brauch’ er zu bitten:²⁹⁴ neg all them need-ø he to beg-inf

‘Oh you all of you divines together, shouted he out, unaware that he did need not to beg all of them.’

(374) Doch nicht brauch’ ich mein Thun vor

euch zu erzählen, but neg need I my deeds in.front.of you to tell-inf Pelasger,/ Mein’ ich; ihr sahet es ja:²⁹⁵ Pelasger think I you saw it part

‘But I do not need to tell you about my deeds, Pelasger, as I think, because you have already seen them. ’

(375) Nicht, daß den

Pöantiden verweilt die vulkanische neg that the-dat Poeantides-dat stays the vulcanic Lemnos,/ Brauch’ ich Rede zu stehn;²⁹⁶ Lemnos need I speech to stand-inf

291 Odyssee, XXIV, 406 translated by Johann Heinrich Voß, (1781). 292 Illias, XIII, 275, translated by Johann Heinrich Voß, (1793). 293 Verwandlungen, Third book – Pentheus, I, 192.104, translated by Johann Heinrich Voß, Berlin: Friedrich Vieweg der Ältere (1798). 294 Verwandlungen, Sixth book – Niobe, I. 330.116, translated by Johann Heinrich Voß, Berlin: Friedrich Vieweg der Ältere (1798). 295 Verwandlungen, Thirteenth book – Ajax and Ulysses, II.284.29, translated by Johann Heinrich Voß, Berlin: Friedrich Vieweg der Ältere (1798).

2.2 Case studies |

189

‘I do not need to justify why the Poeantides maintain the vulcanic Lemnos.’

The pattern in (373) deserves closer attention, as it exhibits a 3rd person singular form which lacks the t-suffix, just like preterite presents do. Yet, given the AcIpattern in (376) taken from Ovid’s Latin original, it is fairly likely that the elided form brauch’ employed by Johann Heinrich Voß has actually to be cosidered as a syncope of an underlying subjunctive of the present form, which is spelled out as brauche. In this type of configuration, it is much more plausible to derive a form as the one given in Voß’ verse. (376) dixerat ignarus

said-3PS unaware-nom

[aci non omens esse rogandos ]²⁹⁷ neg all-acc be-inf ask-ger.acc

‘Unawerly, he said that not all of them are to ask.’

As pointed out by Till Kulawik (pers. commun.), the interpretation of the form brauch’ as a subjunctive of the present is fairly plausible, as this grammatical mood is uniquely used to mark reported speech, just as the AcI-complement does in the Latin original. At any rate, for some mysterious reason the form brauch’ is only used in the first editions of the translation provided by Voß, and is replaced with the less marked form braucht in later editions. The 3rd person singular pattern brauch’ will be discussed in more detail below. As this tendency is reflected by the other examples as well, it appears that brauchen was already restricted to the zu-infinitive in the 18th century. The very rare instances of positive brauchen with an infinitive that are found in that period will be discussed below. The examples in (369) and (371) merit closer inspection, as they involve a recurrent pattern. Both of the instances of brauchen occur in questions that contain a causal use of the interrogative pronoun was ‘what’. As Holler (2009) observes, was is occasionally interpreted as a causal wh-pronoun, much in the way of why. As can be seen, early instances of brauchen with zu-infinitives can often be found in this configuration. Accordingly, it is plausible that this pattern had an impact on the development of brauchen. In the early 18th century, the use of brauchen with an infinitive complement was not addressed in the respective entry of the main dictionaries, cf. Kramer (1702a: 142). This is further confirmed by Scaffidi-Abbate (1973: 4), who provides a comprehensive overview of the main grammars from the 17th and the 18th centuries. Adelung (1793: 1162) is the first one to mention brauchen with zu in a very brief 296 Verwandlungen, Thirteenth book – Ajax and Ulysses, II.309.329, translated by Johann Heinrich Voß, Berlin: Friedrich Vieweg der Ältere (1798). 297 I am indebted to Till Kulawik, who helped me to find and properly analyse the corresponding original Latin example taken from Ovid.

190 | 2 Modal verbs: A class struggle remark. His observations are supported by evidence from two writers from the 18th century: Gellert and Wieland. Becker (1836: 177 §91) has pointed out that brauchen with zu-infinitive belongs to the field of modal verbs. Likewise, Grimm (1837: 949) provides an example exhibiting the IPP. Subsequent grammarians, such as Schoetensack (1856: 295, 297) begin to stress the close relation between brauchen and auxiliaries (Hülfsverb). In what follows, it will be investigated to what extent brauchen shares the essential properties of the traditional six modal verbs. What properties speak in favour of an analysis of brauchen as a modal verb? Three different types of properties will be taken into consideration: morphological features, syntactic features and semantic features. As far as morphological aspects are concerned, brauchen is optionally realised as an irregular form when used in the subjunctive of the past that carries an Umlaut, as is illustrated in (359). However, this irregular form is not restricted to uses of brauchen with an infinitive complement. Kolb (1964: 74), Scaffidi-Abbate (1973: 32) and Girnth (2000: 120) argue that this development is an approximation towards the morphological paradigm of the six traditional modal verbs. They assume that it is an essential characteristic of these verbs that they form the subjunctive of the past with a form containing the Umlaut. Accordingly, they conclude that it is this fact that causes brauchen to chose this unexpected form. However, this is not entirely correct. Firstly, almost every irregular verb in German forms its subjunctive of the past form with an Umlaut. In the canonical case, the indicative past tense form of an irregular verb serves as the base for the subjunctive of the past form in which the stem vowel is altered by an Umlaut, cf. war ‘be-pst’ – wäre ‘be-sbjv.pst’, kam‘come-pst’ – käme ‘come-sbjv.pst’ or wußte ‘know-pst’ – wüßte ‘know-sbjv.pst’. Crucially, this does not hold for two of the core members of the traditional six modal verbs: The indicative past form and the subjunctive past form of wollen and sollen are in each case identical and do not carry any Umlaut: wollte and sollte. Furthermore, the two verbs do not fulfil all of the morphological criteria typical of preterite presents, as illustrated in Section 2.1.1. Alternatively, one could assume that the development of the irregular form of brauchen has pragmatic reasons. Being a regular verb, brauchen does not have distinct forms for the indicative past and the subjunctive of the past form. Any verb in German can alternatively be realised as an analytic subjunctive of the past based on the subjunctive auxiliary würde ‘would’, as was demonstrated in Section 2.2.7. As brauchen already selects an infinitive complement, a further auxiliary would increase the degree of syntactic complexity and processability. For this reason, the development of a proper synthetic subjunctive of the past form for brauchen turns out to be a way which facilitates communication. According to this assumption, the irregular form bräuchte should first have emerged with brauchen with an infinitive complement. Thus, there has to be a period in which bräuchte occurred more

2.2 Case studies |

191

often with infinitives than with accusative NPs. In any case, it remains to be seen to what extent the transitive use of brauchen and its use with a raising pattern have the same preference for the morphological realisation of the subjunctive of the past. It could turn out that one pattern exhibits a significantly stronger preference to be realised in the past subjunctive as the strong form, rather than the weak form. An alternative explanation is based on the observation that the stem brauchcould yield an umlaut in other derivations, such as the plural of the nominalisation Bräuche ‘costums’, or the adjective bräuchlich ‘in use’, ‘usual’. As these forms were very common at this period, speakers may have been inclined to adopt the umlaut for the subjunctive of the past as well. Finally, this morphological peculiarity might have a different reason. Due to ¯ a lack of recorded preterite forms, it is not obvious at all to what extent bruhhan in Old High German belonged to the regular -jan or irregular -an verbs, as has been illustrated by Kolb (1964: 68) and Scaffidi-Abbate (1973: 15ff.). The earliest attested occurrences of bräuchte originate from the 18th century, approximately the same period when brauchen started to select infinitive complements. In both instances, it selects an accusative NP rather than an infinitive complement. This is not surprising, as the transitive use of brauchen has always been the dominant one. Moreover, it already reflects the new interpretation in which it expresses a need for something rather then the use of something. (377)

Bräuchte man hier nicht einen Oedip?²⁹⁸ need-sbjv.pst one here neg a Oedip ‘Do we not need an Oedip here?’

(378)

Ein Fürst bräuchte keine Tugend, wenn er seyn will, a sovereign need-sbjv.pst no virtue if he be-inf wants was er soll, ein Vater des Vaterlandes²⁹⁹ what he should a father the-gen fatherland-gen ‘A sovereign would not need any virtue if he wants to be what he ought to be: A father of the fatherland.’

(379)

/ vnnd mehr Ackers Wo er darüber grieffe where he over touch-sbjv.pst and more acre bräuchte / zu solchem seinem sondern nutz / darfür need-sbjv.pst to such his particular use therefore

298 Johann Jakob Hemmer, Abhandlung über die deutsche Sprache zum Nutzen der Pfalz, Mannheim (1769), p. 221 §111. 299 Franz Berg, Predigten über die Pflichten der höheren und aufgekärten Stände bey den bürgerlichen Unruhen unserer Zeit. Wirzburg: Stahel (1793), p. 54.

192 | 2 Modal verbs: A class struggle

soll er mir als vil frucht in der Schewer geben vnnd bezalen shall he me as much fruit in the barn give-inf and pay-inf / als die selben Ecker vngefehrlich trügen.³⁰⁰ as the same acres approximately bear-sbjv.pst ‘If he uses more acres for his personal purpose, he should give and pay me approximately as much fruit into the barn as the same acres bore.’

Even if these examples do not involve infinitive complements, both occurrences exhibit some parallels to the patterns with infinitive complements. On the one hand, they originate from the same period when brauchen started to take infinitive complements, as will be illustrated below in more detail. On the other hand, both of them occur in a negative environment. As it turns out, the subjunctive form bräuchte can already be found without infinitive complements in the 18th century. According to the earliest occurrences found in the investigation discussed here (cf. 377–378), it is very likely that bräuchte developed independently from the infinitive complement. This is in strict contrast to the development of the IPP-effect: until the present day the IPP is excluded from transitive uses, even in East Bavarian dialects, which are known to be rather generous in the use of the IPP-effect, as has been shown by Aldenhoff (1962: 199) and Ørsnes (2007: 131). This indicates that the morphological peculiarities of brauchen with an infinitive developed in distinct stages. As will be shown below, brauchen can be found with IPP-effect only around 1830. This is also the same period in which the first instances with bare infinitive complements occur. In brief, these facts cast further doubt on the assumption that the exceptional subjunctive form bräuchte has anything to do with the morphological peculiarities of the preterite presents, as has been suggested by Kolb (1964: 74), Scaffidi-Abbate (1973: 32) and Girnth (2000: 120). Their explanation would become even less plausible if example (379) from the 16th century turned out to be valid. As far as the context is concerned, the example is perfect, all of the other lexical verbs bear the subjunctive of the past endings as well: grieffe touch-sbjv.pst and trügen ‘bear-sbjv.pst’. For some mysterious reason, the irregular form bräuchte has been replaced with the regular one brauchte in the subsequent editions of this book from 1574 and 1577. An analogous case can be found in Moscherosch’s Alamodischer Politicus from 1647. Once again, the umlaut is absent from the preceding edition from 1640.³⁰¹

300 Johann Peter Zwengel, New Groß Formular und vollkommlich Cantzlei Buch von den besten und außerlesenen Formularien aller deren Schrifften/ so in Chur und Fürstlichen / auch der Grauen / Herren unnd anderen fürnemen Cantzleyen / Auch sonst in den Ampten unnd Ampts händeln / Deßgleichen under dem gemeinen Man / allerley fürfallendert geschäfft halben / bräuchlich seindt – Reuerßbrieff vber bestandnen Bawhofe/sampt deren Güter Frankfurt: Egenolffs Erben (1568), p. 158a.

2.2 Case studies |

193

As long as the status of the Umlaut of these examples is not confirmed, they will not be taken into further consideration. Yet, it deserves attention that the irregular form bräuchte is subject to criticism: According to Grebe et al. (1966: 114), the irregular subjunctive of the past bräuchte is not correct and has its origins in South German varieties. Similar judgements are found in the Duden dictionary for orthography edited by Scholze-Stubenrecht (2000: 245), in which it is mentioned that bräuchte is characteristic of colloquial language. Finally, the language purist Johannes Dornseiff in his book published in 2011 and entitled Sprache wohin? pleads for replacing the irregular subjunctive of the past bräuchte with the regular form brauchte. One of the main morphological characteristics of preterite present verbs is that they do not carry a suffix in the present tense forms of the 1st and the 3rd person singular, as shown in Section 2.1.1. As Kiaulehn (1965: 52) notices, brauchen has lost its t-suffix in the 3rd person singular whenever it takes a (bare) infinitive complement in spoken German in Berlin. Likewise, Folsom (1968: 328 Fn. 84), Wurzel (1984: 117 & 149) and Birkmann (1987: 5) observe that the t-suffix in the 3rd person singular is about to disappear in other spoken varieties as well. Furthermore, Scaffidi-Abbate (1973: 32) shows that the omission of the t-suffix in the 3rd person singular indicative can be found in the spoken varieties in former Prussia, Rhineland and Palatine. As Girnth (2000) illustrates, the t suffix of the 3rd person singular braucht has also disappeared in many dialects from Rhineland. Furthermore, he argues that this loss of the suffix is due to a morphological process of approximation which brauchen undergoes. Exhibiting modal meaning, it is about to adopt a morphological form for the 3rd person singular that corresponds to that of the six traditional modal verbs and which is spelled out as ‘brauch’. By contrast, Maitz and Tronka (2009: 192) point out that this loss has an independent phonological motivation, as the loss of a dental plosive is also recorded after a velar fricative in other words such as nicht ‘not’ → nich. As Maitz and Tronka (2009: 201) further illustrate, it is not excluded that this process may have been amplified by the morphological factors discussed by Girnth (2000). In any case, the loss of the suffix -t in the third person singular is restricted to the spoken language of particular varieties. The case of need in English exhibits a striking parallel. It was already shown by Sweet (1891: 421) that need optionally drops the suffix -s in the 3rd person singular indicative, preferably when used with an infinitive complement. Murray (1933: 71) could find this pattern in texts from the 16th century onwards. This leads

301 Johann Michael Moscherosch, Alamodischer Politicus, Cölln: Andreas Bingen (1640), p. 64. Johann Michael Moscherosch, Alamodischer Politicus, Cölln: Andreas Bingen (1647), p. 56.

194 | 2 Modal verbs: A class struggle

Scaffidi-Abbate (1973: 32) to assume explicitly that the loss of the t-suffix in the case of brauchen is to be considered as a process of morphological approximation toward the paradigm of preterite present verbs. Another essential morphological characteristic of the six traditional modal verbs is the so-called infinitivus pro participio (IPP-)effect, which was discussed in great detail in Section 2.1.1.2. As was shown, there is only a small group of verbs to which the IPP has to apply obligatorily, among them the traditional six modal verbs. As Kolb (1964: 76) and Schmid (2000: 331) point out, brauchen exhibits an obligatory IPP whenever it used with an infinitive complement. In the corpus examples (380)–(381), the IPP form brauchen cannot be substituted with the ge-participle gebraucht. (380)

Faust hätte die beiden ja auch einfach gar nicht zu Faust have-sbjv.pst the both part also just intn neg to erwähnen brauchen.³⁰² mention-inf need-ppp(ipp) ‘It was not necessary that Faust mentioned both of them.’

(381)

Da habe Blaich nur zuzugreifen brauchen.³⁰³ there have-sbjv.prs Blaich only to.take-inf need-ppp(ipp) ‘In this situation, Blaich just needed to take it.’

(382)

die Regierung hätte das Buch gar nicht zu verbieten the government have-sbjv.pst the book intn neg to prohibit-inf brauchen, es wäre dennoch gelesen worden.³⁰⁴ need-ppp(ipp) it is-sbjv.pst nevertheless read-ppp pass.aux-inf ‘It was not necessary that the government prohibited the book, it would have been read nevertheless.’

(383)

Nun eine solche hätte es auch gerade nicht zu seyn so a such have-sbjv.pst it also just neg to be-inf brauchen;³⁰⁵ need-ppp(ipp) ‘So, it was not necessary’

302 DeReKo: RHZ09/MAI.19683 Rhein-Zeitung, 23/05/2009. 303 DeReKo: BRZ07/MAI.18723 Braunschweiger Zeitung, 18/05/2007. 304 Heinrich Heine, Reise nach Italien, Abschnitt IV, in Morgenblatt für Gebildete Stände, N° 20, Mittwoch 3. December 1828, (1828) p. 1157. 305 Carl Ullmann, Theologische Studien und Kritiken, volume II (1) Hamburg: Friedrich Perthes (1829), p. 176.

2.2 Case studies |

(384)

195

dem Soldaten hätt ich nicht einmal brauchen einen the soldier have-sbjv.pst I neg once need-inf a Schlaftrunk zu geben, er wäre doch nicht aufgewacht.³⁰⁶ sleeping.draught to give-inf he sbjv-pst yet neg wake.up-ppp ‘It was not necessary to give the soldier a sleeping draught, he would not have woken up anyway.’

The earliest attested examples of brauchen exhibiting an IPP are from the early 19th century (cf. 382 and 383), the former from Sanders (1908: 101). At the same time, Grimm (1837: 949) acknowledges that brauchen with zu-infinitive is found with the IPP-effect. His example is discussed in Section 2.1.1.2. In the section on the IPP, Grimm (1837: 168) declares that any verb that selects a zu-infinitive complement must not be combined with IPP-morphology. Interestingly, the Grimm brothers nevertheless employ brauchen with an IPP in their collection of fairy tales (cf. 384). By contrast, the normative philologist Wustmann (1908: 61) argues that brauchen with ge-participle is fehlerhaft ‘wrong’ and ungeschickt ‘awkward’ as soon as it selects a zu-infinitive complement. Aldenhoff (1962: 196) and Kolb (1964: 77) report that a whole range of grammarians in the 18th century share Grimm’s opinion. As they furthermore argue, this view is almost correct; however, there is one verb that does not conform to this tendency: brauchen, which is frequently found with zu-infinitives in this pattern. Likewise, Askedal (1997a: 55) demonstrates that brauchen with zu-infinitive does not fulfil all of the criteria required by the configuration in which the IPP usually occurs. Today, it is generally acknowledged that brauchen can occur with the IPP despite its unexpected type of infinitive complement. How does this strange situation come about? As is evident, the driving force behind it are the fact that brauchen gradually adopted the semantic properties müssen and the existence of the ancient pattern of dürfen. Interestingly, the IPP was not the only option for brauchen with zu-infinitive. Numerous instances of brauchen with zu-infinitives can be found that are realised as ge-participles. The earliest dates back to the late 18th century (cf. 385), the latest originates from the late 19th century (cf. 389). Similar examples have been discussed by Sanders (1908: 101).

306 Jacob Grimm and Wilhelm Grimm, Die zertanzten Schuhe in Kinder und Hausmärchen. Große Ausgabe, (1837), p.553.

196 | 2 Modal verbs: A class struggle

(385)

So unsäglich reich, daß er gleich 200.000 fl so inexpressibly rich that he immediately 200.000 fl wegschenken kann, hätte vielleicht Herr Ifler auch nicht away.give-inf can have-sbjv.pst maybe Mister Ifler also neg zu seyn gebraucht.³⁰⁷ to be-inf need-ppp(ge) ‘Mister Ifler did not need to be that inexpressibly rich that he could give away 200.000 fl at once.’

(386)

So hätte z. B. die Beschreibung des Zimmers in der so have-sbjv.pst e. g. the description the-gen room-gen in the Abtey, in welches Cordelia gebracht wird, für den Zweck abbey in which Cordelia bring-ppp pass.aux for the purpose des Dichters weit weniger umständlich zu seyn the-gen poet-gen far less cumbersome to be-inf gebraucht.³⁰⁸ need-ppp(ge) ‘For example, the description of the room in the abbey into which Cordelia was brought did not need to be as cumbersome for the purpose of the poet.’

(387)

Man sieht sehr oft diese Pflicht ganz unrichtig an; als ob man one sees very often these duties very incorrect on as if one verbunden sein könnte, dieses oder jenes, was man außerdem obliged be-inf could this or that what one besides nicht zu tun gebraucht hätte (etwa in die Kirche, neg to do-inf need-ppp(ge) have-sbjv.pst about in the church zum Abendmahle gehen u. dgl.), um des bloßen to.the supper go and the.like for the-gen mere-gen guten Beispieles zu tun.³⁰⁹ good-gen example-gen to do ‘One very often deliberates about this duty in a wrong manner, as if one could be obliged to do something which was not necessary besides to do (e.g to go to church or the supper) just for the sake of being a good example.’

(388)

er hätte nur die Regungen der eigenen Brust zu he have-sbjv.pst only the emotions the-gen own chest to besingen gebraucht³¹⁰ sing need-ppp(ipp)

307 Allgemeine Literatur Zeitung 57 20. Februar 1789, p. 456. 308 Neue Bibliothek der schönen Wissenschaften und der freyen Künste volume 53/1: Adalbert der Wilde by Friedrich August Müller. Leipzig Dyckische Buchhandlung (1794), p. 281. 309 Johann Gottlieb Fichte, Das System der Sittenlehre volume II, Leipzig Meiner (1798), p. 716.

2.2 Case studies |

197

‘He only needed to sing about the emotions in his chest.’

(389)

Eine blendende Art der Darstellung wäre mir a splendid kind the-gen description-gen be-sbjv.pst me wahrscheinlich gelungen, wenn nur nichts hinter derselben zu probably succed-ppp if only nothing after it to seyn gebraucht hätte,³¹¹ be-inf need-ppp(ge) have-sbjv.pst ‘I would have managed to deliver a splendid kind of description if it had not been necessary to write something after it.’

All of the occurrences listed above involve circumstantial modality. This does not seem to be obvious. The first two examples (385) and (386) are taken both from a review of a theatre play or novel, in which the reviewer criticises certain details of the work. In the investigation discussed here, brauchen with an infinitive complement was found with a ge-participle in 1789, whereas the earliest occurrence of the variant with an IPP-effect dates back no earlier than 1829. This could lead us to the conclusion that the ge-participle was the original form which was replaced with the IPP in the course of the 19th century. Furthermore, it appears that brauchen with an infinitive complement was initially restricted to zu-infinitives. The first example of brauchen with a bare infinitive complement was mentioned by Sanders (1908: 101) in the late 19th century (cf. 390): (390)

ich hätt’ mich bloß nicht einmischen brauchen³¹² I have-scjv.pst refl only neg barge.in-inf need-ppp(ipp) ‘It was not necessary that I barged in.’

His example is taken from a novella written by Paul Heyse, which talks about the life in the mountains around the Watzmann in Southern Bavaria. This occurrence of brauchen is part of a longer report told by the character Sepp, which is intended to reflect the spoken language of the local people. However, it certainly does not correspond to the actual language use in this region. Nevertheless, this example indicates that brauchen with bare infinitive complement was considered as a feature of spoken language. The use of brauchen with a bare infinitive complement is not mentioned in grammars until the late 19th century. On the one hand, Wustmann (1896: 57), in

310 Heine 2, 307, [as cited in Sanders (1908, 101)]. 311 Edmund Jörg und Franz Binder, Historisch-politische Blt¨ter für das katholische Deutschland volume 26, München (1870), p. 237. 312 Paul Heyse, Auf der Alm in: Neue Novellen volume 4, (pp.385–455) Berlin: Wilhelm Hertz (1862), p. 447.

198 | 2 Modal verbs: A class struggle

a derogatory way, considers brauchen without zu as gemeines Berliner Zeitungsdeutsch ‘Common Newspaper German of Berlin’. In a later edition, Wustmann (1908: 61) chooses an even more drastic terminology: Gassendeutsch ‘StreetGerman’. On the other hand, Sanders (1908: 101) observes in a less subjective way that brauchen is occasionally used with a bare infinitive. Based on data from the late 19th century, he notices that brauchen can occur with a bare infinitive complement, in particular in spoken language. Interestingly, the example given above (390) involves an IPP which according to some grammarians at that period should only be used with verbs that take bare infinitive complements. Until today, the IPP is a property which predominantly occurs with verbs that select a bare infinitive complement. According to the scarce data presented here, the suspicion arises that the emergence of bare infinitive complements with brauchen is a consequence of the increasing use of the IPP. Obviously, the ability to realise an IPP acts as a threshold between auxiliary-like verbs and lesser grammaticalised verbs: once a verb with zu-infinitive develops a meaning that is close to the core-class of IPP verbs, as defined by Schmid (2000: 328), it is likely to acquire the IPP-effect as well. As soon as it has adopted this new property, there are two options. Either it drops the infinitival particle zu and begins to subcategorise bare infinitive complements, such as brauchen, or it loses the ability to realise its participle as IPP, as is exemplified in the case of many other verbs that used to exhibit the IPP for a certain period such as wissen or pflegen. A more comprehensive collection of verbs which is are observed with the IPP-effect is provided by Sanders (1908: 222) and Aldenhoff (1962), as is illustrated in Section 2.1.1.2. In any case, all of these speculations require a systematic survey of a significant amount of data from the 18th and the 19th centuries. The absence of brauchen with bare infinitives in the data discussed could be due to the selection of texts. It may turn out that it was already used much earlier, or that is was only used in the spoken language. Moreover, it is conceivable that grammarians prior to Sanders (1908: 101) and Wustmann (1908: 61) ignored brauchen with bare infinitive because it was considered non-standard language. Yet, the latter conclusion is not very plausible, as Langer (2001) has demonstrated that grammarians of the 18th and the 19th centuries had a particular interest to fight against “bad language”. Accordingly, they were well informed about recurring non-conform uses of language. Turning to the syntactic characteristics, it has been mentioned that the six traditional modal verbs take bare infinitive complements. Some grammarians, such as Jäger (1968: 332), only consider the use of brauchen with zu-infinitive complements acceptable. By contrast, Folsom (1968: 328) argues that brauchen without zu is fully acceptable. Other grammarians, such as Sanders (1908: 101) recommend dropping the infinitive particle zu in certain environments: (i) whenever

2.2 Case studies |

199

the infinitive complement is topicalised, and (ii) in cases in which brauchen itself is realised as a zu-infinitive, it is sufficient to use one single infinitive particle zu. This recommendation corresponds roughly to the one made by the popular Duden grammar edited by Drosdowski et al. (1995: 395 Fn.1), which is based on the claim made by the same author in an earlier study, cf. Gelhaus (1969: 320).³¹³ Likewise, the highly respected Austrian writer Karl Kraus argued in the early 20th century that brauchen without zu should not be considered as incorrect. Furthermore, he notices that this option is even more appropriate in environments with more than one infinitive particle zu.³¹⁴ Turning to the investigation of the DeReKo corpus, it turns out that brauchen occurs with bare infinitives even in written language. Further instances have been found in works of fiction of high renown, such as the pattern (398) taken from Franz Kafka’s Verwandlung (‘The Metamorphosis’). Note that none of the examples (391)–(401) refers to spoken language or reported speech, except example (392) taken from Götzke and Mitka (1939: 411). The occurrences can be divided into four classes, according to the type of environments in which they occur. First of all, (i) brauchen with bare infinitive complements can be used if its infinitive complement is topicalised (cf. 391–394). Secondly, (ii) the particle zu is occasionally omitted in contexts in which the phonetic string zu occurs adjacent to the infinitive complement such as allzu ‘too’ (cf. 394) or zur ‘to.the’ (cf. 396) or zugeben ‘admit’ (cf. 401). Furthermore, (iii) brauchen without zu can be found in verb clusters that consist of at least three verbs (cf. 395–398). And finally, (iv) there are instances of brauchen with bare infinitive complements that exhibit the IPP (cf. 398–401). In some cases, these factors bundle: Topicalisation and multiple

313 An extract from the original quotation: Aus gründen des Wohlklangs (. . . zu tun haben statt . . . zu tun zu haben) und in der Emphase (Wundern braucht man sich nicht! wird brauchen auch standardsprachlich ohne zu vorgezogen) English translation [J.M.]: brauchen is preferred in standard language without zu for reasons of euphony (. . . zu tun haben instead of . . . zu tun zu haben) and in topicalisations (Wundern braucht man sich nicht! instead of brauchen). 314 Karl Kraus Die Fackel, März 1925 München: Kösel Verlag 2nd edition, p.265–266. The full quotation is as follows: An und für sich ist brauchen ohne zu keineswegs falsch – das wurde nie behauptet –, es gibt Fälle, in denen es sogar vorzuziehen ist, eben wenn sich die zu häufen oder wo eine mehr mundartliche Färbung oder Veranschaulichung des abgekürzten Vorgangs intendiert ist.

200 | 2 Modal verbs: A class struggle

zu (cf. 394), clusters of verbs and IPP (cf. 398 and 400), multiple zu and IPP (cf. 401). As most of these examples involve a certain degree of morphological and syntactic complexity, most native speakers would not realise that these patterns do not correspond to the rules of normative grammars. In order to identify the influence of these four environments, they should be investigated with respect to other verbs that select zu infinitive complements. It appears that such verbs do not drop the zu-particle that easily. (391)

Wandern braucht ihr nicht zum Süden, weil ihn ihr wander-inf need you neg to.the south because him her Gesang euch bringt.³¹⁵ song you brings ‘You do not need to wander to the South, as her song will bring it to you.’

(392)

Ja, totschlagen brauch’ ich ihn nicht, er verdient’s gar yes strike.dead-inf need I him neg he merits.it intn nicht³¹⁶ neg ‘Yes, I do not need to strike him dead, he does not merit it.’

(393)

Politiker müssen sich Kritik gefallen lassen, beschimpfen politicians must refl criticism please-inf let-inf insult-inf oder beleidigen lassen brauchen sie sich nicht.³¹⁷ or offend-inf let-inf need they refl neg ‘Politicians have to be able to deal with criticism, but they do not need to acquiesce to insults and offenses.’

(394)

Allzu sehr grämen brauchen sich die Viertäler angesichts der all.too much worry-inf need refl the Viertäler regarding the Tabellenlage allerdings nicht.³¹⁸ table.position however neg ‘However, the team from Viertal do not need to worry regarding their position in the table.’

315 Friedrich Rückert, Nachklang in Gedichte (1841), p. 299. First published in Morgenblatt für gebildete Stände N° 243, Mittwoch 10. October 1821 (1821). 316 Willibald Alexis, Der Werwolf Berlin: Otto Janke, (1894) part I, chapter 9, p.89. First published in 1848. 317 DeReKo: RHZ07/FEB.00452 Rhein-Zeitung, 01/02/2007. 318 DeReKo: RHZ08/MAR.20777 Rhein-Zeitung, 25/03/2008.

2.2 Case studies |

(395)

201

Sei großmütig, da die Strafe in Deiner Hand liegt und be-imp noble as the punishment in your hand lies and Du nur mich lange auf einen Antwortsbrief warten lassen you only me long for a answer.letter wait-inf let-inf brauchst.³¹⁹ need ‘Be noble, as the punishment lies in your hands and you only need to keep me waiting a long time for an answer.’

(396)

Aber was sicherlich nicht zur Diskussion gestellt werden but what certainly neg to.the discussion put-ppp pass.aux-inf braucht, ist die Tatsache, daß viele Dichter und Schriftsteller needs is the fact that many poets and writers vom 19. Jahrhundert bis zum heutigen Tag das Verb from.the 19 century until to.the present day the verb ‘brauchen’ mit reinem Infinitiv anwenden.³²⁰ need with bare infinitive use ‘Yet, what certainly does not need to be discussed is the fact that many poets and writer from the 19th century until the present day use the verb ‘brauchen’ with a bare infinitive complement.’

(397)

Die Erkundungen ergaben, dass zwei Stollen nicht verfüllt the explorations yielded that two adits neg fill-ppp werden brauchten.³²¹ pass.aux-inf needed ‘The explorations have yielded that two adits did not need to be closed.’

(398)

[. . . ] so hätte noch keineswegs die Mutter so have-sbjv.pst still by.no.means the mother for für sie eintreten müssen und Gregor her advocate-inf must-ppp(ipp) and Gregor have-sbjv.pst hätte doch nicht vernachlässigt werden brauchen.³²² part neg neglect-ppp pass.aux-inf need-ppp(ipp) ‘Accordingly, the mother would not have by no means to have advocate her and Gregor would not have needed to be neglected.’

(399)

Dabei hätte es soweit nicht kommen brauchen.³²³ though have-sbjv.pst it so.far neg come-inf need-ppp(ipp)

319 Droysen’s letter to Heydemann 20th June 1840, as cited in Götzke and Mitka (1939: 411). 320 August Scaffidi-Abbate ‘Brauchen’ mit folgendem Infinitiv, in: Muttersprache 83 (1973), p. 5 321 DeReKo: BRZ07/JAN.17470 Braunschweiger Zeitung, 03/01/2007. 322 Franz Kafka, Die Verwandlung, (1916) p. 179 l. 3.

202 | 2 Modal verbs: A class struggle

‘Though, it was not necessary that this happened.’

(400)

Eigentlich hätten die Aachener Bosch gar nicht gehen actually have-sbjv.pst the Aachenian Bosch intn neg go-inf lassen brauchen, denn sein dortiger Vertrag sah eine let-inf need-ppp(ipp) as his local contract envisage a Auflösungsklausel frühestens für 2014 vor.³²⁴ cancellation earliest for 2014 before ‘Actually, there was no need for the Aachenians to let Bosch go as his local contract did not envisage a cancellation before the year 2014.’

(401)

Dass er seine Bütt’ erst am Vortag geschrieben hat, that he his vat.speech only at.the previous.day written has hätte er nicht offen zugeben brauchen.³²⁵ have-sbjv.pst he neg publicly admit-inf need-ppp(ipp) ‘It was not necessary that he admitted that he only wrote his polemic on the previous day.’

Interestingly, most of the occurrences were found in newspapers from Eastern Austria and Rhineland. Pfeffer (1973: 92) already noticed that brauchen with bare infinitives is frequent in the spoken language of Austria and absent from the one in Switzerland. Based on some earlier data collected by Pfeffer, Kluempers (1997) comes to a similar conclusion. While these uses can be easily found for the 20th century, it is not obvious when this pattern emerged. Sanders (1908: 101), Götzke and Mitka (1939: 411) and Scaffidi-Abbate (1973: 5) provide a couple of occurrences taken from literature of the 19th century. However, a number of them are taken from direct speech or dialogues that should reflect the use of the spoken language, such as (392). The earliest examples of brauchen in genuine written language that select bare infinitive complements are from the early 19th century. In the Trübners Deutsches Wörterbuch, Götzke and Mitka (1939: 411) provide two interesting occurrences: one involving a topicalisation (cf. 391) of 1821 and another one that contains a verb cluster (cf. 395) of 1840. The claim put forth by Götzke and Mitka (1939: 411) that brauchen without zu can be already found in the collection of tales edited by the Brothers Grimm could not be confirmed with the support of electronic corpora: such forms could neither be found in the version provided by the DeReKo corpus nor in the version accessible in the Digitale Bibliothek Deutscher Klassiker.³²⁶

323 324 325 326

DeReKo: RHZ10/APR.12082 Rhein-Zeitung, 22/04/2010. DeReKo: NUN10/JUN.01746 Nürnberger Nachrichten, 18/06/2010 DeReKo: NON10/JAN.05921 Niederösterreichische Nachrichten, 18/01/2010 http://klassiker.chadwyck.co.uk/deutsch/home/home, last access 9th July 2012.

2.2 Case studies |

203

These examples show that there are environments in which a native speaker would not notice that brauchen is used without the infinitive particle zu. By contrast, a number of speakers would not consider brauchen with a bare infinitive complement as prestige Standard German, as soon as it occurs in other environments than those four discussed above. As has been illustrated on various occasions, brauchen has mostly adopted patterns that its predecessor dürfen had already established. In this respect, it is fairly astonishing that dürfen can be found with zu-infinitives in its old interpretation in the early 16th century, as Scaffidi-Abbate (1973: 26) demonstrated, referring to examples provided by the DWB (p. 1722), (cf. 402). This is fairly unexpected, as all of the preterite present verbs already selected bare infinitive complements for several centuries. Does this mean that there are independent semantic reasons that prescribe the selection of the zu-infinitive? (402)

nun sprichstu waz darf ich das zu biten, so ich doch daz on now talk.you what need I that to beg-inf as I yet that and daz hab, dan ich bin reich³²⁷ that have since I am rich ‘Now you talk, what necessitates me to beg for that? As I have this and that, as I am rich.’ ‘Now you talk, why do I need to beg for that? As I have this and that, as I am rich.’

Note that the configuration in example (402) is reminiscent of the ones in (369) and (371): It contains a causal use of the wh-pronoun waz and a verb that expresses a need. Similar examples with causal was and thurfan are already documented for the Old High German Tatian, as the example provided by Scaffidi-Abbate (1973: 25) illustrates. (403)

Uuaz thurfun uuir noh nu urcundono³²⁸ what need we yet now witnesses ‘Why do we now yet need witnesses?’

This observation could be the crucial hint for an alternative explanation for the fact that dürfen could occur next to zu-infinitives. Apart from the selection of bare infinitive complements, brauchen exhibits a further characteristic typical of the six traditional modal verbs. Just like volitional mögen, raising wollen and dürfen in the Early New High German period, brauchen with an infinitive is restricted to negative environments. This has often been observed. Crucially, as Paul (1897: 79) already pointed out, this property also affected

327 Geiler Keiserberg sünden des munds, 84a (1518), as cited in the DWB. 328 Tatian, 191,2 (830) as cited in Scaffidi-Abbate (1973: 25).

204 | 2 Modal verbs: A class struggle

the new transitive uses of brauchen. As soon as transitive brauchen referred to a need rather than a usage, it was restricted to negative polar contexts. This is of particular interest, as the transitive use of the need related interpretation lost this property in the course of time, whereas brauchen with an infinitive complement continues to avoid positive environments. Bech (1951: 14), Kolb (1964: 73) and Lenz (1996: 399) have demonstrated that brauchen in most respect adopted the function of the original need verb dürfen, including its negative polar orientation. Folsom (1968: 325) provides a comprehensive list of items that license a suitable negative polar environment for brauchen: explicit negation, particles that imply some sort of negation nur ‘only’, kaum ‘hardly’ and erst ‘(temporal) only’, subordinative conjunctions ohne dass ‘without that’ and polarity and wh-questions. However, there are rare instances of brauchen that do not occur in negative polar environments. First of all, the negative polarity of brauchen poses a particular challenge for language learners. Due to overgeneralisation, they sometimes assume that brauchen can also be used in positive contexts, as the examples produced by a L1-learner (404) and by the Turkish L2-learner Sevinc (405) provided by Vainikka and Young-Scholten (1994: 289) illustrate. Kürschner (1983: 193) discusses more positive examples of brauchen from L1-acquisition. (404)

doch Papa das brauchst Du kaufen.³²⁹ part daddy that need you buy-inf ‘Oh yes, daddy you need to buy this.’

(405)

Jetzt brau Wohnungsamt fragen.³³⁰ Now need housing.authority ask-inf ‘Now, I need to ask the housing authority.’

Folsom (1968: 326) and Pfeffer (1973: 88) have moreover collected relevant instances produced by native speakers of German. Whereas Folsom provides examples from the 19th century, Pfeffer discusses an example from Contemporary Spoken German (cf. 406). As Roland Schäfer (pers. commun.) has pointed out to me, brauchen is occasionally used in positive environments (cf. 407–408). According to his evaluation, this appears to be a regional phenomenon of the spoken language in North Rhine-Westphalia. (406)

Es brauchen immer so die sehr schwierigen Sachen zu sein³³¹ it need always so the very difficult things to be-inf ‘It always has to be the very difficult things.’

329 Utterance heard in a bookstore in Berlin in the year 2000, when a small child argued with his father. 330 As quoted in Vainikka and Young-Scholten (1994: 289). 331 As quoted in Pfeffer (1973: 88).

2.2 Case studies |

(407)

205

Jeder Kinositzer braucht ein eigenes Bild berechnet zu each cinema.sitter needs a proper picture calculate-ppp to bekommen, da die Abstände vom Bild der einzelnen get-inf as the distances from.the picture the-gen single-gen Besucher viel zu gravierend sind.³³² visitor-gen much too serious are ‘It is necessary that each cinema goer gets an individually calculated picture as the distances from the picture of each visitor are too varied.’

(408)

Jetzt brauch ich meine Pillen nehmen, und dann geh ich ins now need I my pills take and then go I in.the Büro.³³³ office ‘Now, I just need to take my pills and than I’ll go to the office.’

There is evidence that related patterns were already common in the 18th century. However, as this pattern is acceptable for speakers who do not consider examples (ex:BrauchenSachenZuSein)–(ex:BrauchPillenNehmen) as grammatical, it appears that the environment obviously still has properties of an negative polarity context in German, possibly due to the presence of the wh-pronoun in the zero relative clause. (409)

so weiß der Patriach, was er zu wissen braucht, mehr als er so knows the patriarch what he to know needs more than he braucht.³³⁴ needs ‘As a consequence, the patriarch knows what he needs to know, even more than he needs.’

Since Folsom and Pfeffer published their investigations, brauchen does not seem to have undergone any dramatic change in Standard German. Even if there was evidence that brauchen could spread to positive environments as well, it remained confined to negative polar distribution until the present day. And this is one of the big mysteries: Why did the new transitive use of brauchen lose its negative polar behaviour, whereas brauchen with an infinitive is still confined to negative environments after two hundred years? 332 Eternal thanks to Roland Schäfer, who provided me this example found on 26th of January 2011 on the web: http://forum.golem.de/kommentare/audio-video/roger-ebert-us-filmkritiker-erklaert-den-3dfilm-fuer-tot/was-3d-vor-allem-fehlt/48624,2629841,2629841,read.html#msg-2629841 333 Eternal thanks to Roland Schäfer, who supplied me with this example produced by a 70 year old speaker from Southwestern Phalia. on 28th of March 2011. 334 Gotthold Eprhaim Lessing, Nathan der Weise, 1. Aufzug, 5. Auftritt (1779).

206 | 2 Modal verbs: A class struggle

Apart from the selection of bare infinitives and the preference for negative polar environments, brauchen has another property that is typical of the six traditional modal verbs. As Folsom (1968: 324) and Askedal (1997a: 57) have demonstrated, brauchen can select VP-anaphora that refer to infinitives which are embedded by a traditional modal verb. Such patterns are well documented in the DeReKo corpus (cf. 410–411). (410)

Das heißt, man kann eine direkte Patenschaft übernehmen, that means one can a direct sponsorship overtake-inf braucht es aber nicht?³³⁵ need vpana but neg ‘Does this mean that one can adopt a sponsorship, but one does not need to?’

(411)

Dass Chelseas Spieler-Trainer Gianluca Vialli seinem Captain that Chelsea-gen player-manager Gianluca Vialli his Captain selbst diesmal Nachsicht versprochen hat, müsste eigentlich even this.time clemency promised have should actually verwundern. Braucht es aber nicht. Vialli und Wise sind surprise-inf need vpan but neg Vialli and Wise are Freunde.³³⁶ friends ‘It should actually surprise us that the playing manager Gianluca Vialli was even lenient towards his captain this time. However, it doesn’t need to. Vialli and Wise are friends.’

As it turns out, circumstantial brauchen behaves exactly like müssen with respect to negation. Accordingly, it exhibits all of the essential raising diagnostics that were introduced in Section 2.2.1.3 and 2.2.2.3. In the corpus, it is found with the non-referential subject es ‘it’ (cf. 412–413) and with de dicto interpretations of quantifying NPs (cf.414). It should not be relevant here whether eines is used as an indefinite pronoun or as a numerical determiner, as Carpenter (1998: 87) has pointed out that numerical determiners behave like ordinary existential quantifiers. Finally, the subject referent in example (415) is not the target on which the obligation is imposed, as we would expect for a control pattern. (412)

Es braucht nicht überall Weisswürste und Bier oder it need neg everywhere white.sausage and beer or «Züri-Gschnätzlets» zu geben.³³⁷ Züri-Gschnätzlets to give-inf

335 DeReKo: M00/DEZ.79640 Mannheimer Morgen, 20/12/2000. 336 DeReKo: 99/FEB.03243 Zürcher Tagesanzeiger, 05/02/1999.

2.2 Case studies |

207

‘It isn’t necessary that Weisswurst, beer or Züri-Gschnätzlets is available everywhere.’

Vor dem Winter braucht es den Eichhörnchen nicht bang before the winter need it the-dat squirrel-dat neg afraid zu werden.³³⁸ to become-inf

(413)

‘Before winter, the squirrel doesn’t need to be afraid.’

Von jeweils zwei Geschwistern braucht nur eines Eintritt zu of each two siblings need only one entrance.fee to bezahlen.³³⁹ pay-inf

(414)

‘Only one out of two siblings has to pay the entrance fee.’

Bei Grippe muss unter allen Umständen das Bett gehütet with influenza must under all circumstances the bed tend-ppp werden – es braucht nicht das eigene zu sein.³⁴⁰ pass.aux-inf it need neg the own to be-inf

(415)

‘In case of influenza, it is necessary to stay in bed under any circumstance – it does not need to be the own one.’

Finally, raising predicates are transparent with respect to voice. If brauchen was a control verb, the target of the obligation would be expected to always be the subject referent, as any control verb determines the main semantic relations by means of assigning thematic roles. Involving different subjects, the sentences (416a) and (416b) should accordingly refer to different states of affairs. However, they differ: In the active example (416a) the subject is the underlying agent argument der Reinhold, whereas the subject is identified with the underlying theme argument der Nanga Parbat in the passivised instances. If brauchen involved a proper subject argument, the obligation should be imposed on Reinhold in (416a) and on the Nanga Parbat in the passivised example (416b). This is not the case, so brauchen must involve a raising pattern. (416)

a.

Der Reinhold braucht den Nanga Parbat nicht ohne the Reinhold must the-acc Nanga Parbat neg without Sauerstoffgerät (zu) bezwingen. oxygen.apparatus to conquer ‘Reinhold has to conquer the Nanga Parbat without oxygen mask.’

337 DeReKo: E00/JAN.01750 Zürcher Tagesanzeiger, 20/01/2000. 338 DeReKo: RHZ96/NOV.14124 Rhein-Zeitung, 22/11/1996. 339 DeReKo: NON09/MAR.11321 Niederösterreichische Nachrichten, 18/03/2009. 340 DeReKo: NUZ06/FEB.03100 Nürnberger Zeitung, 27/02/2006.

208 | 2 Modal verbs: A class struggle

b.

Der Nanga Parbat braucht vom Reinhold nicht ohne the-nom Nanga Parbat must by.the Reinhold neg without Sauerstoffgerät bezwungen (zu) werden. oxygen.apparatus conquer-ppp to pass.aux-inf ‘The Nanga Parbat has to be conquered by Reinhold without oxygen mask.’

Just as müssen in the scope of negation, negative polar brauchen can act as a quantifier over individuals, yielding a quantificational interpretation in the sense of Carlson (1977: 119) and Brennan (1993: 96). In example (417), nicht müssen serves as a negated universal quantifier over individuals (¬∀). It expresses that in the set of people with cancer there is at least one person whose fate is not hopeless. The following examples (418)–(423) behave in parallel way. (417)

Ein Leben mit Krebs braucht nicht hoffnungslos zu sein.³⁴¹ a life with cancer need neg hopeless to be-inf ‘A life with cancer does not need to be hopeless.’

(418)

Was umweltfreundlich ist, braucht noch lange nicht what eco.friendly is need still long neg menschenfreundlich zu sein.³⁴² charitable to be-inf ‘What is ecologically friendly, does not need to be charitable.’

(419)

Und der Prager Komponist Petr Eben hat vorgeführt, daß Musik and the Prager composer Petr Eben has demonstrated that music im Riesenraum nicht zwangsläufig plakativ zu sein in.the giant.space neg inevitably blatant to be-inf braucht.³⁴³ needs ‘And Prague’s Composer Petr Eben has demonstrated that music in the giant space does not need to be blatant. ’

(420)

Nicht jeder, der vielleicht ein guter Lebenshelfer ist, braucht neg everyone who maybe a good life.helper is needs auch ein guter Schriftsteller zu sein³⁴⁴ also a good writer to be-inf ‘Not everybody who is maybe good at giving advise need to be a good writer.’

341 342 343 344

DeReKo: A98/MAR.13892 St. Galler Tagblatt, 06/03/1998. DeReKo: N93/MAR.07932 Salzburger Nachrichten, 03/03/1993. DeReKo:N93/OKT.39664 Salzburger Nachrichten, 30/10/1993. A00/APR.27086 St. Galler Tagblatt, 20/04/2000.

2.2 Case studies |

(421)

209

Teurere Produkte brauchen nicht zwangsläufig besser zu sein, expensive products need neg inevitably better to be-inf aber oft ist der höhere Preis auch ein Zeichen grösserer but often is the higher price also a sign bigger-gen Qualität und eines besseren Services.³⁴⁵ quality-gen and a-gen better-gen service-gen ‘Expensive products do not need to be inevitably better, yet, a higher price often indicates a higher quality and a better service.’

(422)

bei dieser Gelegenheit hatte er, sowohl mit sich selbst als mit at this occasion had he both with refl self as with Serlo und Aurelien, die Frage oft abgehandelt, welch ein Serlo and Aurelia-dat the question often dealt which a Unterschied sich zwischen einem edlen und vornehmen circumstance refl between a noble and distinguished Betragen zeige, und inwiefern jenes in diesem, dieses aber conduct shows and to.what.extent this in that that yet nicht in jenem enthalten zu sein brauche.³⁴⁶ neg in this contain-ppp to be-inf need-sbjv.prs ‘On this occasion, he had often discussed the question with himself on the one side and Serlo and Aurelien on the other side to what extent there is a difference.’

(423)

In der Sprache der Kinder und der in the language the- children-gen and the-gen Geisteskranken erscheint ein Inf., der nicht mentally.ill.persons-gen appears an infinitive that neg elliptisch zu sein braucht, als allgemeine Form, die die elliptically to be-inf needs as general form that the anstrengende Ausprägung bestimmter logischer Beziehugen exhausting specification certain-gen logical-gen relation-gen nicht erfordert.³⁴⁷ neg requires. ‘In the language of children and mentally ill persons, an infinitive appears that does not need to involve an ellipsis, being a general form that does not require the exhausting specification of certain logical relations.’

345 DeReKo:97/JUN.09921 St. Galler Tagblatt, 17/06/1997. 346 DeReKo: GOE/AGM.00000 Goethe: Wilhelm Meisters Lehrjahre, [Roman], (Erstv. 1795–1796), p. 352. 347 Behaghel, Otto Deutsche Syntax II, (1924) p. 364.

210 | 2 Modal verbs: A class struggle As example (422), taken from the 18th century, shows, the quantificational interpretations of brauchen seem to have already been available right after the period when it had acquired the infinitive complement. 2.2.9.5 Raising directionals with event modification As was shown by Welke (1965: 15), Folsom (1968: 323), Kürschner (1983: 194), Askedal (1997a: 57) and Szumlakowski (2010), brauchen is also found with verbless directional phrases. In this respect, it behaves exactly like müssen. As both examples involve inanimate subjects, an analysis in terms of ellipsis does not seem to be very likely, as was illustrated in Section 2.2.1. (424)

Das Fahrrad braucht nicht zum TÜV.³⁴⁸ the bike need neg to.the TÜV ‘The bike does not need to be sent to the Technical Control Board’

(425)

Dieser Wagen paßt auf Anhieb, er braucht nicht in die the car fit at first.go he need neg in the Änderungsschneiderei.³⁴⁹ alteration.tailor ‘This car fits at first go, it doesn’t need to be send to the alteration tailor.’

Like its counter part with an infinitive complement, brauchen continues to be a negative polarity item in this use. In this regard, it very much resembles the negative polar uses of the raising pattern of wollen and the volitional use of mögen, which continue to be negative polar whenever they are employed with a verbless directional phrase complement. 2.2.9.6 Subject-to-object raising infinitives At least in some varieties such as Viennese, brauchen can be used as a subjectto-object raising verb. In this use it imposes strict selectional restrictions on the infinitive, which has to be a stative locative verb such as liegen ‘lie’, sitzen ‘sit’ or stehen ‘stand’. In that respect, it is very similar to other subject-to-object raising verbs such as haben and the relinquative use lassen, and the subject-to-subject raising verb bleiben, as has been illustrated by Maché and Abraham (2011: 260): (426)

I brauch di da jetzt net deppat umanand sitzn. I need you-acc there now neg stupid around sit-inf ‘It doesn’t help me if you sit around here now.’

348 DeReKo: RHZ09/JAN.18261 Rhein-Zeitung, 24/01/2009. 349 DeReKo:RHZ97/AUG.13358 Rhein-Zeitung, 23/08/1997.

2.2 Case studies |

211

Interestingly, the subject-to-object raising pattern is negative polar just like the subject-to-subject raising pattern with zu-infinitive is. 2.2.9.7 Raising infinitives with clause modification It is a matter of debate to what extent brauchen can exhibit an epistemic interpretation. On the one hand, some authors, e.g. Folsom (1968: 328), Kürschner (1983: 192) and Öhlschläger (1989: 8), contest that this type of interpretation is acceptable. On the other hand, Ehlich and Rehbein (1972: 340), Raynaud (1977: 22), Takahaši (1984: 21), Zifonun (1997: 1278), Askedal (1997a: 62) and Kluempers (1997: 101) provide examples from corpora which they consider to be epistemic. At the same time, most of them concede that this type of interpretation occurs fairly rarely with brauchen, unlike Takahaši (1984: 21). In the past decades, a couple of instances of brauchen with an infinitive complement have been collected that are candidates for an epistemic interpretation. Takahaši (1984: 21) discusses (427), Vater (2010: 108) considers brauchen in (428) as epistemic and Folsom (1968: 323) provides (429) and (430). Interestingly, he does not notice that they can plausibly be interpreted in an epistemic way. It was only Takahaši (1984: 21) who noticed that they can be interpreted epistemically. Likewise, Askedal (1997a: 62) remarks that Folsom’s example (429) is most likely to be an epistemic instance of brauchen. (427)

Das braucht nicht der Fall zu sein.³⁵⁰ that need part negthe case to be-inf ‘That doesn’t really have to be the case.’

(428)

Das braucht nicht zu stimmen. this need neg to be.right-inf ‘That doesn’t have to be correct.’

(429)

Es braucht nicht unbedingt ein Irrtum des Computers it need neg necessarily a error the-gen computer-gen gewesen zu sein. be-ppp to be-inf ‘It does not need to have been an error by the computer.’

(430)

Ebenso braucht nicht unbedingt sie die Eintrittskarte in Iesolo likewise need neg necessarily she the ticket in Iesolo verloren zu haben. lose-ppp to have-inf ‘Likewise, it does not need to be her who has lost the ticket in Iesolo.’

350 H. Gipper, Sprachwissenschaftliche Grundbegriffe und Forschungsrichtungen, 1978.

212 | 2 Modal verbs: A class struggle

Do the instances given above indeed involve epistemic modal operators? As far as their distribution is concerned, these examples behave like typical epistemic modal verbs. Such instances of brauchen embed predicates that refer to states that cannot be changed (cf. 427–428) and predicates that refer to events in the past (cf. 429–430), just as genuine epistemic modal verbs do. Moreover, brauchen in the examples above is subject to the CoDeC: The speaker labels the embedded proposition as a proposition that is not part of his confirmed knowledge. Thus, they are not compatible with the assertion of the same proposition. As the investigation of the DeReKo corpus has revealed, the number of corpus examples for epistemic brauchen is fairly limited. The retrieval strategy used is as follows. In her own small corpus study investigating 6000 modal verbs, Raynaud (1977: 22) found out that 90 % of the epistemic modal verbs embed the stative predicate sein. Furthermore, Heine (1995: 23) demonstrates that infinitive perfect complements strongly favour an epistemic interpretation. According to the frequencies documented by Ruoff (1981), the stative predicates sein and haben make up together almost 50 % of the verbs used in his corpus. Thus, if brauchen does not occur in an epistemic interpretation with any of these stative predicates, it is not very likely that it will occur as an epistemic modal verb at all.³⁵¹ The study presented here has focused on instances of brauchen that co-occur with the negation nicht and the verbs sein or haben. Both verbs can either occur as stative predicates or perfect tense auxiliaries which is of course part of the infinitive perfect. The first query for brauchen that selects the complement zu haben yielded 768 hits. Far more than 80 % of these occurrences were variations of the pattern nicht Angst zu haben brauchen ‘it is not necessary to be afraid’, in which the noun Angst can be replaced with a semantically related noun such as Sorge ‘worry’, Befürchtungen ‘fear’, Respekt ‘respect’, Ekel ‘disgust’, Scheu ‘timidity’, schlechtes Gewissen ‘bad conscience’. Of course, all of these examples do not come into consideration for an epistemic interpretation. In the most typical case, they are employed as direct or indirect advise. Surprisingly, there are only two occurrences out of 768 that display a clear epistemic interpretation: (431) and (432). Apart from them, there are just a couple of cases that could turn out to be epistemic as well. Yet, the contexts are too ambiguous to determine the precise interpretation.

351 The investigation has been carried out on April 22nd 2010. The first query &brauchen /s0 ¨nicht¨ /s0 zu haben yielded valid 768 results and the second query &brauchen /s0 ¨nicht¨ /s0 zu sein yielded valid 1683 results. Of course, this query does not extract all of the negative polar interpretations of brauchen. But as nicht is by far the most frequent negative operator, this study has ignored the lesser frequent ones for the sake of simplicity.

2.2 Case studies |

(431)

213

Was den Ort Xanten als Ort der Sage betrifft, so ist what the village Xanten as location the-gen myth concerns so is Norbert Lönnendonker der Auffassung, dass das Santen Norbert Lönnendonker the-gen opinion that the Santen des Nibelungenliedes nicht am Niederrhein gelegen zu the-gen Nibelungenlied-gen neg at.the Lower.Rhine lie-ppp to haben braucht³⁵² have-inf needs ‘As for Xanten as the location of the myth, Norbert Lönnendonker believes that the village Santen appearing in the Nibelungenlied was not necessarily located at the Lower Rhine.’

(432)

Wir haben die Telekom längst gebeten, vor Ort We have the Telekom long.ago asked at place nachzusehen. Das braucht Herr Kunz gar nicht gemerkt zu after.to.look-inf this need Mister Kunz intn neg notice-ppp to haben, weil der Techniker dafür nicht unbedingt have-inf because the technician therefore neg necessarily ins Haus muss³⁵³ into.the house must ‘We already asked the Telekom company to check his connection long ago. Mister Kunz may not necessarily noticed it because the technician does not need to enter the house to do so.’

As regards the other predicate that was investigated, sein, the situation is not much different. As the study has revealed, there are only a couple of occurrences out of 1683 that can be taken into consideration for epistemic interpretation. And in a number of instances, it is hard to determine whether brauchen exhibits an epistemic or quantificational reading. (433)

gäbe es keine im weiten Sinn wirtschaftlichen, give-sbjv.pst it no in.the broader sense economically international verträglichen, umweltverträglichen und internationally reconcilable biocompatible and sozialverträglichen Technologien, so wäre die social.compatible technologies so be-sbjv.pst the industriegesellschaftliche Entwicklung in eine Sackgasse industry.social development into a blind.alley

352 DeReKo:WPD/SSS.10575, Wikipedia – URL:http://de.wikipedia.org: Wikipedia, 2005. 353 DeReKo:NUN06/NOV.02580 Nürnberger Nachrichten, 23/11/2006.

214 | 2 Modal verbs: A class struggle

geraten. Dies aber braucht unseres Erachtens nicht der Fall zu got this but need our consideration neg the case to sein.³⁵⁴ be-inf ‘If there were no technologies that were economically and internationally reconcilable, biocompatible and socially compatible in the broader sense, the development of the industrial society would have reached a blind alley. According to our estimation, this does not need to be the case.’

(434)

Der „Outer”, wie man klar erblickt, ist nunmehr regelrecht the outer as one clearly sees is now downright verrückt. Denn was er sagt, gesteht er ein, braucht überhaupt crazy as what he says confesses he in need at.all nicht wahr zu sein³⁵⁵ neg true to be-inf ‘The outer has gone downright crazy, as one can clearly see. Since what he says doesn’t need to be true at all anymore; as he confesses.’

It is fairly surprising that epistemic brauchen almost never occurs in environments in which epistemic modal verbs are usually more frequent than circumstantial modal verbs. This unexpected behaviour calls for an explanation. The instances of epistemic brauchen here differ from canonical epistemic modal verbs in two respects: (i) they select zu-infinitive complements rather than bare infinitive complements, and (ii) they are restricted to negative environments. According to Reis (2001: 307, 312 Fn. 39, 2005: 112), the canonical type of infinitive complements for epistemic verbs in German is the bare infinitive. Thus, she would expect that zuinfinitives inhibit an epistemic interpretation. This could explain why so few occurrences could only be found in the corpus. Accordingly, another investigation needs to be conducted in which brauchen selects bare infinitive complements. As they do not occur in written language so frequently, this is not a simple task. Alternatively, the low frequency of an epistemic interpretation could be caused by the presence of a negation. As will be discussed in more detail in Section 4.10, a number of authors consider the scope of negation an environment in which an epistemic operator cannot occur. In a similar manner Askedal (1991: 9, 1997: 62), argues that this context is not a suitable context in which brauchen can develop an epistemic interpretation without restrictions. In order to determine which of these approaches is the correct one, it is fruitful to take a look at müssen in the scope of negation. In this section, it was seen

354 DeReKo:H86/UZ3. 20139 Die Zeit, 28.03.1986. 355 DeReKo:O95/SEP.86494 Neue Kronen-Zeitung, In den Wind gereimt, 02/09/1995.

2.2 Case studies |

215

that the two necessity verbs with negation behave almost identically in semantic terms. In case it should turn out that epistemic müssen in the scope of negation is equally rare with the stative complements sein and haben, the suspicion arises that it is the negation that inhibits the occurrence of epistemic interpretations. In case epistemic nicht müssen occurs considerably more frequently under these conditions, the infinitival particle zu can be held responsible for the low frequency of epistemic (nicht) brauchen. Summing up, the position defended by Takahaši (1984), Askedal (1997a) and Reis (2001) can, by and large, be confirmed: brauchen is documented with an epistemic interpretation. However, the number of occurrences is much smaller than is expected. Accordingly, brauchen can be considered a verb with a marginally developed epistemic interpretation. Finally , it will be shown that the epistemic reading of brauchen is not a recent development. There are occurrences as early as the late 19th century, as is indicated in (435): (435)

Aber es wird nun ein Mahl als ausgemacht angenommen, das But it pass.aux now a time as agreed assumed the Ganze, woraus die sogenannten Fragmente sind, ensemble where.from the so-called fragments are habe nur ein einziges Buch betragen, und zwar have-sbjv.prs only a single book amount and indeed habe es kein Größeres zu seyn gebraucht, als das have-sbjv.prs it no bigger to inf need-ppp(ge) as the zweyte Buch von Gajus.³⁵⁶ second book by Gajus ‘It is taken for granted that the ensemble from which the so-called fragments originate only made up a single book and it does not need to have been more voluminous than Gajus’ second book.’

There is yet another candidate for an epistemic interpretation which occurs in the 18th century. Despite the fact that this sentence could have been ambiguous for speakers at that period, the more likely interpretation of brauchen in example (436) is a teleological one: In order to survive the cold climate, the tree did not have to be too contrarious. In an epistemic interpretation, the author would rather express an assumption about the nature of the tree.

356 Gustav Hugo Beyträge zur civilistischen Bücherkenntnis der letzten vierzig Jahre Berlin: August Mylius (1829), p. 646.

216 | 2 Modal verbs: A class struggle

(436)

Es läßt sich als ausgemacht annehmen, daß die edleren it let refl as agreed assume-inf that the precious Obstsorten, welche niemals wild wachsend gefunden, sondern fruits which never wild growing found but allezeit unter menschlicher Wartung und Pflege erzeugt werden always under human attention and care produced are von einer gemeinen und wilden Mutter herstammen, welche from a common and feral mother stem which nachmals durch die Länge der Zeit, mit Hülfe later through the length the-gen time-gen with help menschlichen Nachdenkens, Kunst und Fleißes, ihre Zucht human-gen reasoning-gen, art-gen and effort-gen their growth veredelt und an Figur, Farbe, Geschmack, Geruch und Größe cultivated and on shape colour taste smell and size verändert hat. Dieser Mutterbaum, obgleich wild, hat doch, wenn changed has this mother.tree, even.if feral has yet if er unter einem milden Luftstriche stand, nicht so herbe und he under a mild air.flow stood neg so harsh and widrig zu seyn gebraucht, wie die Aepfel, welche unsre contrarious to be-inf need-ppp(ge) as the apples which our nordischen Waelder erzeugen.³⁵⁷ Nordic forests produce ‘It is commonly assumed that the fruit trees which have never been found feral in nature but which were always produced under human attention and care originate from a common and feral mother. Consecutively, they have been cultivated their growth supported by human reasoning, art and effort. Thus, they changed their shape, colour, taste, smell and size. This mother tree didn’t need to be as harsh and contrarious as the apples that grow in our Nordic forests (epistemic: This mother tree needn’t have been as harsh and contrarious as the apples that grow in our Nordic forests).’

These occurrences merit closer attention for various reasons: Firstly, they involve an epistemic modal verb that is realised as a past participle. As will be shown in Section 4.2, such environments for an epistemic modal verb may seem awkward from the perspective of a contemporary native speaker of German. Most authors do not accept such patterns. However, in Spanish and French this pattern is the most natural one. Accordingly, we could assume that epistemic past participles were possible in earlier stages of German as well. And, indeed, it will be shown in Section 4.2 that there are other verbs that could be employed in an analogous way.

357 Peter Jonas Bergius Von Obstgärten und deren Beförderung in Schweden Leipzig: Gräffische Buchhandlung (1794), p. 40.

2.2 Case studies |

217

Secondly, the past participle is still realised as a ge-infinitive, rather than an IPP. This illustrates that the form does not play the major role in the development of an epistemic interpretation. Finally, it is interesting that brauchen had developed an epistemic variant as early as two centuries ago, and that its frequency has not increased until the present day. 2.2.9.8 Brief sketch of the development of brauchen Based on the scarce evidence that could be found for the present investigation, the following scenario is the most likely one for the historical development of brauchen. The development can be divided into four phases. Phase I: until 1650. As has been illustrated by Adelung (1793: 1162), Paul (1897: 79), Kolb (1964: 65) and Scaffidi-Abbate (1973: 5), the transitive verb brauchen used to refer to the usage of something until the 17th century. Then, it acquired a new interpretation and could express a need for something. In both variants, the verb selected either a genitive NP or an accusative NP. Crucially, the new variant of transitive brauchen started out as a negative polarity item, as has been illustrated by Paul (1897: 79). According to Scaffidi-Abbate (1973: 5) and Reis (2005a: 106), the new meaning of brauchen was the crucial condition that facilitated the selection of infinitive complements. Phase II: 1650–1700. brauchen acquires an impersonal pattern. It is likely that this development was caused by language contact with the French pattern il faut ‘it is necessary . . . ’. Furthermore, it is plausible to assume that the availability of a variant without referential subject argument facilitated the development of a raising pattern. Grimm and Grimm (1860: 318) provide an example in the DWB that could reveal the precise development of brauchen (cf. 437). (437)

es braucht nun gezeigt zu werden, dasz dieser prinz einer it needs now show-ppp to pass.aux-inf that this prince a solchen abscheulichkeit fähig war.³⁵⁸ such hideousness able was ‘It is only necessary to demonstrate that this Prince was capable of committing such a hideousness.’

It is worth mentioning that brauchen in the example given above is not a negative polarity item. Furthermore, it subcategorises the non-referential subject es. These facts indicate that this example most probably involves the impersonal pattern of brauchen. Whereas the object argument is usually realised by a genitive or accus-

358 Friedrich von Schiller: Werke II 54 Band Geschichte des dreissigjährigen kriegs, zweyter teil, Carlsruhe Bureau deutscher Klassiker (first edition 1792) 1823), p 137.

218 | 2 Modal verbs: A class struggle

ative NP, it possibly surfaces here as an infinitive complement. This assumption receives additional support, as Grimm and Grimm (1860: 318) give a another example of impersonal brauchen that selects a finite clause as its object argument instead of a NP. Likewise, Kluempers (1997: 21, 87) considers the impersonal use of brauchen as the starting point of its grammaticalisation, even if he assumes that this use is already an instance of subjectification. Phase III: 1700–1800. In the early 18th century, brauchen can be found for the first time with a zu-infinitive complement. In this phase, brauchen with an infinitive realises its past participle as a ge-participle. In the same period, the irregular subjunctive of the past form bräuchte can already be found. As can be seen, brauchen had already developed the full range of functions by the end of the 18th century: It could be used as a quantificational modal verb and as an epistemic modal verb. It is at the end of phase III that brauchen with zu-infinitive was mentioned by a grammarian for the first time, namely Adelung (1793: 1162). As Kolb (1964: 75), Welke (1965: 75) and Scaffidi-Abbate (1973: 24) point out, müssen is not frequently used with a matrix negation – accordingly, another element became necessary that was suitable to fill the gap left by müssen. This observation is in conflict with the corpus study conducted by Brünner (1979: 82). In her corpus of spoken language, she found 525 occurrences of müssen without a negation, and 20 occurrences in the scope of negation. As regards nicht brauchen, it was found 27 times. According to her point of view, negated müssen occurs most commonly in relation with können or müssen without negation. Phase IV: 1800 – present. Finally, brauchen started to realise its past participle as an IPP. At this point, brauchen cannot be found with a bare infinitive yet. Grimm (1837: 168, 949) notices that the IPP is a property that is restricted to verbs that take bare infinitives. This should thus apply to brauchen, which selected zu-infinitives. At this time, the first uses of brauchen with a bare infinitive occur. In the late 19th century, the normative philologist Wustmann (1908: 61) rejects this as incorrect German. Moreover, it is found in the 3rd person singular without a suffix, which could potentially be interpreted as a present indicative (rather than subjunctive) form, cf. the discussion of example (373). As there were similar verbs in neighbouring languages that were already more grammaticalised before brauchen started its development, it is likely that the development of brauchen is a result of language contact. Murray (1933: 71) illustrates that need, the English counterpart of brauchen, developed into an modal auxiliarylike verb much earlier. First of all, it could be used in an impersonal pattern as early as the 14th century. Moreover, need is already documented from the late 14th century on with to-infinitives, and from the late 15th century with bare infinitive complements. In the 16th century, it started to lose the s-suffix in the 3rd person

2.2 Case studies |

219

singular. Some of these developments are illustrated by the examples taken from Murray (1933: 71). A similar observation has been made by Denison (1993: 296). (438)

How prejudicial such proceedings are . . . need not be defined.³⁵⁹

(439)

My stooping need not to have disturbed you. footnotemark

The emergence of negative polar need in late Middle English appears to be related to the development of the necessity modal verb must. This contemporary use of that latter verb merits attention for two reasons: In morphological respect, must is the former past form of the possibility verb mote, which acquired a present meaning, subsequently developing a paradigm of its own. In this respect, it has undergone a development that is fairly similar to the one of the preterite presents and the former subjunctive of the past möchte. Secondly, it has changed its modal force in the course of its development. Whereas mote originally referred to a possibility, the new form must can only be employed as a necessity modal verb in Modern English. Thirdly, it displays an idiosyncratic behaviour with respect to the scope of negation. In contrast to the other modal verbs of English, a negated form must not is interpreted with a narrow scope interpretation (□¬p). Accordingly, the negation of a necessity (¬□p) cannot be expressed by a modal auxiliary in English. This, finally, explains why there is a gap in the paradigm for a new specialised modal verb: need not, which is used to negate a necessity. This scenario is supported by the data provided by Murray (1933: 791). The new modal auxiliary must with present meaning can be found from the 14th century onwards. The old possibility readings of mote were in use until the 15th century, whereas the first necessity readings developed in the 11th century. The narrow scope interpretation of must not is at least documented until the 16th century. A similar view is held by Kaita (2012: 394). The role of influence of foreign languages is, once again, highlighted by a pattern familiar from French. The verb faillir ‘need’ is found with an impersonal pattern from the 14th century onwards. This make an influence from foreign languages very plausible. Kluempers (1997) provides an alternative path of development, but as he only grounds his theory on a couple of examples taken from the DWB, his account lacks empirical justification.

359 Richard Morris An essay in defence of ancient architecture p. 90, (1728), as cited in Murray (1933: 71). 359 Thomas Hull The story of Sir William Harrington II, 9 (1771), as cited in Murray (1933: 71).

220 | 2 Modal verbs: A class struggle

2.2.10 werden The first author to suggest a modal analysis for werden is Vater (1975), although Bauer (1870: 157 §164 Fn.) already argued in a footnote that future tense can have modal meanings as well, illustrating this claim with an epistemic use of werden. He is inclined to consider werden as marker for a (future) probability. In contrast, Vater (1975: 110) has noticed that werden exhibits an epistemic interpretation that is analogous to the one that is typical of the traditional modal verbs. Likewise, the English counterpart will exhibits an epistemic reading as well. Accordingly, Brennan (1993: 97) and Enç (1996: 356) have adopted a modal account for the English future auxiliary, which can be analysed as a necessity modal. As Bres and Labeau (2012) illustrate, this holds for the French future auxiliary aller ‘go’ with bare infinitive complement as well. In contrast to the six traditional modal verbs, werden does not exhibit any interpretation that could be considered as a circumstantial modal one. Accordingly, Öhlschläger (1989: 8) and Reis (2001: 312) do not regard it as a full member of the class of modal verbs. Instead, werden involves two other auxiliary-like functions. First of all, it can be used as a copula with an ingressive or inchoative interpretation taking predicative complements. Moreover, it functions as a passive auxiliary that selects past participle complements. Furthermore, it is used as future auxiliary with bare infinitive complements. Finally, it can be employed as an epistemic modal verb that selects bare infinitive complements as well. 2.2.10.1 Predicative phrases As Steinitz (1999: 145) and Eisenberg (2004: 85) illustrate, werden ‘get’, ‘become’ is used as a copula verb which is similar to sein ‘be’, but specified for an ingressive or inchoative meaning. It takes a predicative complement that is typically realised as an adjective such as kühl ‘cool’ (cf. 440), or as a NP such as ein echter Test ‘a true test’ (cf. 441). Moreover, it can be realised as a directional phrase such as aus dem Einbrecher ‘out of the burglar’ or zum Paradies or ‘into the paradise’ (cf. 444) – this relativises Maienborn’s (1994: 232) generalisation, according to which directional phrases are generally no licit complement for copulas. Apart from that, it turns out that the copula werden can select non-referential subject NPs. This indicates that it is a raising predicate. Similar observations for other copula constructions have been made by Pollard and Sag (1994: 147) and Müller (2002: 72, 2009: 217, 2013: 7):

2.2 Case studies |

(440)

221

In der kommenden Nacht wird es mit 11 bis 9 Grad in the upcoming night gets it with 11 to 9 degrees empfindlich kühl.³⁶⁰ sensitively cool ‘In the next night, the temperatures will decrease to a level of 11 or 9 degrees’

(441)

Diese Wahl wird ein echter Test für die große Koalition.³⁶¹ this election becomes a true test for the big coalition ‘This election turns into a true test for a big coalition.’

(442)

So wird aus dem Einbrecher plötzlich ein Bigamist.³⁶² so gets out the burglar suddenly a bigamist ‘Suddenly, the burglar turns into a bigamist.’

(443)

Strohwein ist eine edelsüße Spezialität, die in Deutschland straw.wine is a noble.sweet speciality that in Germany rar geworden ist.³⁶³ rare become-ppp(ge) is ‘Straw wine is a sweet speciality which became rare in Germany.’

(444)

Spanien ist zum Paradies der Schönheitschirurgie in Europa Spain is to.the paradise the-gen aesthetic.surgery in Europe geworden.³⁶⁴ become-ppp(ge) ‘Spain became a paradise of aesthetic surgery in Europe.’

Embedded by the perfect auxiliary sein, the copula werden is realised as the geparticiple geworden, as is illustrated in (443) and (444). 2.2.10.2 Passive auxiliary According to Müller (2002: 147, 2007: 306), the passive auxiliary werden is most efficiently analysed as a verb that takes a past participle as its infinitive complement. The crucial property of past participles is that their designated subject argument is not part of their subcategorisation frame anymore. Accordingly, the most prominent argument of the past participle is their object argument. The object argument, in turn, is selected by the passive auxiliary werden and becomes its

360 DeReKo: RHZ04/JUL.26792 Rhein-Zeitung, 29/07/2004. 361 DeReKo: NUN06/DEZ.02328 Nürnberger Nachrichten, 20/12/2006. 362 DeReKo: RHZ09/NOV.07144 Rhein-Zeitung, 09/11/2009. 363 DeReKo: RHZ07/AUG.15619 Rhein-Zeitung, 17/08/2007. 364 DeReKo: NUZ05/DEZ.02372 Nürnberger Zeitung, 20/12/2005.

222 | 2 Modal verbs: A class struggle

syntactic subject. Roughly speaking, the passive auxiliary involves some sort of object-to-subject raising operation. In (445), werden embeds the transitive verb betreiben ‘run’. As a matrix verb, it realises its agent argument as a subject NP with nominative case and its theme argument as an object NP with accusative case. As soon as it is selected by the passive auxiliary werden, it appears as the past participle form betrieben and its designate subject argument will be no longer available in its subcategorisation frame. As a consequence, the theme argument Möbelcenter oder Baumärkte ‘furniture centres or building centres’ raises to the subject position of the matrix predicate werden, in which it is realised as a nominative NP. (445)

Möbelcenter oder Baumärkte werden meist von furniture.centres or building.centres pass.aux most by denselben Konzernen betrieben.³⁶⁵ the.same concern run-ppp ‘Furniture centres or building centres are usually run by the same concern.’

(446)

Der neue Bereich wird von Marcel Klaus geleitet, der the new area pass.aux by Marcel Klaus direct-ppp, who vom Tamedia-Konzern zur Crossair zurückgekehrt ist.³⁶⁶ from.the Tamedia-concern to.the Crossair return-ppp is ‘The new section is directed by Marcel Klaus, who has returned from the TamediaConcernt to the Crossair.’

(447)

Die erste große Universalbank ist 1810 von Rothschild the first bog universal.bank perf.aux 1810 by Rothschild gegründet worden.³⁶⁷ found-ppp pass.aux-ppp ‘The first big universal bank has been founded in 1810 by Rothschild.’

(448)

Nach mehr als drei Wochen ist die deutsche Archäologin after more than three weeks is the German archaeologist Susanne Osthoff von ihren Entführern frei gelassen Susanne Osthoff by her kidnappers free let-ppp(ge) geworden.³⁶⁸ pass.aux-ppp(ge) ‘After, three weeks the German archaeologist Susanne Osthoff has been released by her kidnappers.’

365 366 367 368

DeReKo: R98/JAN.06668 Frankfurter Rundschau, 27/01/1998. DeReKo: A01/NOV.39871 St. Galler Tagblatt, 02/11/2001. DeReKo: R99/FEB.08474 Frankfurter Rundschau, 02/02/1999. DeReKo: BRZ05/DEZ.11398 Braunschweiger Zeitung, 19/12/2005.

2.2 Case studies |

223

According to most standard grammars of German such as Eisenberg (2004: 124) and Eisenberg et al. (2005: 466), werden has an irregular past participle that is reminiscent of the IPP. It is realised as a form without a ge-prefix that involves an alternation of the stem vowel. The past participle is spelled out as worden, as is indicated in (448). Surprisingly, a few hundreds of instances could be found in the archive T of the DeReKo corpus in which the past participle of the passive auxiliary is spelt out as geworden, which is actually the form that belongs to the copula interpretation of werden.³⁶⁹ According to traditional grammars, this option should not exist. Aside from the patterns described above, werden can select verbs which do not carry an accusative object. Lacking a direct object, the result of the passivisation with these verbs is a structure which does not contain a syntactic subject. Traditionally it is called unpersönliches Passiv ‘impersonal passive’, as is discussed by Müller (2002: 118, 2007:220): (449)

Dort wird Menschen geholfen, die sich keinen there pass.aux-sg people-dat.pl help-ppp rel.prn refl no Anwalt leisten können.³⁷⁰ advocate afford-inf can ‘At this place, people get aid which cannot afford an advocate.’

Crucially, the sentence lacks a nominative subject NP. The passive auxiliary exhibits default inflection, which is 3rd person singular in German. The dative object of the passivised verb helfen ‘help’ continues to bear dative case and does not agree with the finite passive auxiliary wird. 2.2.10.3 Future related interpretations There are instances of werden with an infinitive complement in which the embedded predicate refers to an event in the future. It is contested to what extent these uses can be analysed as future tense auxiliary. On the one hand, some authors, such as Krämer (2005: 26), show that this future related interpretation of werden can be clearly distinguished from its epistemic counterpart. Therefore, she assumes that werden has an independent interpretation as a future auxiliary. By contrast, Vater (1975) and Erb (2001: 176) argue that werden with an infinitive complement always has to be considered as a modal auxiliary, even in its uses with future related interpretations. Vater (1975: 95) notices that werden behaves atypic-

369 Investigation carried out on 22nd July 2012. The regular past participle of the passive auxiliary worden could be found 500,000 in the same archive. 370 DeReKo: NUN06/AUG.00493 Nürnberger Nachrichten, 04.08.2006.

224 | 2 Modal verbs: A class struggle

ally, as it can embed complements that refer to the past. As Erb (2001: 146) illustrates, werden with an infinitive occurs in an environment that is similar to modal verbs, rather than to the one in which the perfect tense auxiliary occurs. While Vater (1975: 119) assumes that werden can be either interpreted in a circumstantial modal or an epistemic modal way, Erb (2001: 175) argues that werden is always an epistemic modal verb. Likewise, Enç (1996: 351) has demonstrated that English will differs essentially from genuine tense auxiliaries with respect to sequence of tense effects and embedding other tense operators. Furthermore, authors who advocate a modal analysis of the future auxiliaries werden and will do not agree with respect to the specification of the relevant modal force. Whereas Brennan (1993: 97) and Enç (1996: 356) conclude that the English future auxiliary will is a necessity modal operator, Vater (1975: 113) argues that werden carries a modal force situated between the one of können (possibility) and the one of müssen (necessity). Likewise, Zifonun (1997: 1910) classifies werden as a modal probability operator. Less explicitly, Kratzer (1981: 58) is inclined to treat werden as a necessity modal operator, as she associates it with the adverb certainly. Finally, Kissine (2008) argues for English that will always has to be considered as a future auxiliary, rather than a modal auxiliary. At this point, it is not clear to what extent this line of reasoning could be plausibly defended for German. As it turns out, it is no trivial matter to capture the future oriented uses of werden. But it seems to be useful to draw a careful distinction between the future oriented reading and the epistemic one. In contrast to the analysis suggested by Erb (2001: 175), there are instances of werden that occur in environments in which genuine epistemic modal verbs cannot occur. If all of the instances of werden were indeed epistemic, additional explanations would become necessary to explain the acceptability of the examples below (450)–(453). (450)

Wir werden jetzt 1000 Unternehmen anschreiben und anfragen, we will now 1000 enterprises write-inf and ask-inf ob sie sich an dieser tollen Aktion beteiligen.³⁷¹ whether they refl at this amazing action participate ‘Now, we will write to 1000 enterprises in order to ask whether they want to participate in this amazing action.’

(451)

In der Goldbäckerei Schulze wird es von 11 bis 12 Uhr und in the Goldbäckerei Schulze will it from 11 until 12 o’clock and von 12 bis 13 Uhr wieder eine Kinderbetreuung geben.³⁷² from 12 until 13 o’clock again a child.care give-inf

371 DeReKo: BRZ06/JUN.06668 Braunschweiger Zeitung, 14/06/2006.

2.2 Case studies |

225

‘There will be a child care again in the Goldbäckerei Schulze from 11 am to 12 am and from 12 am to 1.pm.’

(452)

Die Freie Bürgerliste hat allen Grund zum Feiern. Sie wird in the Freie Bürgerliste has all reason to.the celebrate she will bin Zukunft mit drei Mandaten im Gemeinderat vertreten future with three mandates in.the city.council represent-ppp sein.³⁷³ be-inf ‘The Freie Bürgerliste has every reason to celebrate. They will be represented in the city council by three mandates.’

(453)

Er selbst wird dann zwar bereits das 65. Lebensjahr he himself will then part already the 65th live.age überschritten haben.³⁷⁴ pass-ppp have-inf ‘Then, he will already have passed the 65th year of his life.’

First of all, the future oriented use of werden fairly often selects subjects that are specified for the 1st person singular or plural (cf. 450). This is unexpected for any approach that assumes that werden is always an epistemic modal verb: As Heine (1995: 24) has pointed out, the 1st person subjects are very unusual for epistemic modal verbs. Erb (2001: 176) argues that it is the 1st person feature that brings about the future resonance. However, the scope of her explanation is limited and would not apply to any of the other examples given above. Furthermore, Krämer (2005: 23) points out that werden with an infinitive (i) can be embedded in restrictive relative clauses and without any complication, (ii) it can bear verum focus and finally, (iii) it can be embedded by predicates of desire such as hoffen. Crucially, in all of these environments it is restricted to future interpretation. In none of the examples illustrated above is an epistemic interpretation plausible. Assuming that epistemic werden involves a (weak) necessity operator, it should be possible to substitute it with another epistemic necessity operator such as müssen. However, any replacement would drastically decrease the speaker’s commitment to the validity of the proposition. This indicates that there are at least two interpretations of werden with respect to the degree of the speaker’s commitment. Likewise, Krämer (2005: 60) argues that epistemic werden is specified for [- evidence]. This indicates that the commitment to the embedded proposition

372 DeReKo: BRZ07/DEZ. 19912 Braunschweiger Zeitung, 07/12/2007. 373 DeReKo: BVZ07/OKT.01215 Burgenländische Volkszeitung, 10/10/2007. 374 DeReKo: HAZ08/NOV.03088 Hannoversche Allgemeine, 18/11/2008.

226 | 2 Modal verbs: A class struggle

is not based on direct evidence. Such a restriction does not apply to the future interpretations listed above. As it turns out, werden exhibits a whole range of striking analogies with respect to circumstantial modal verbs: It selects a bare infinitive complement, it tolerates non-referential subjects (cf. 451) and it locates the embedded predication in a time interval after the Time of Utterance. However, there are some crucial differences as well. First of all, Erb (2001: 146) illustrates that the future reading of werden is neither found as an infinitive nor as a past participle. This does not apply to circumstantial modal verbs in German. Moreover, English will can be used as a quantificational necessity modal verb, as has been pointed out by Brennan (1993: 97) and Enç (1996: 356). This does not seem to be the case for its German counterpart werden to the same extent. Once again, this illustrates that there are differences between the future reading of werden and the six traditional modal verbs. All of the observations made above indicate that under closer scrutiny, werden can neither be regarded as forming a natural class with the six traditional circumstantial modal verbs, nor with the epistemic modal verbs. 2.2.10.4 Raising infinitives with clause modification As Bauer (1870: 157 §164 Fn.), Welke (1965: 12), Vater (1975: 110), Fritz (1991: 43), Engel (1996: 463) Erb (2001: 161) and Krämer (2005: 57) have pointed out, werden with an infinitive complement can be interpreted in an epistemic way, analogous to the epistemic interpretation of the traditional modal verbs. As Krämer (2005: 128) demonstrates, the epistemic pattern has emerged out of the future interpretation. According to Fritz (1991: 43), this development must have occurred prior to the early 17th century, the period when the first instances of epistemic werden are documented. Just as the traditional modal verbs do, werden is restricted to an epistemic interpretation whenever it embeds a predication consisting of an identified individual and a predicate that refers to a state that cannot be changed, or a predicate that refers to some past event (cf. 456 and 458). (454)

Der Leser wird wohl den wackeren Komponisten Gluck the reader will probably the brave composer Gluck kennen, vornamens Christoph Willibald, einen know-inf, with.first.name Christoph Willibald a Oberpfälzer.³⁷⁵ upper.Palatinate

2.2 Case studies |

227

‘The reader will probably know the composer Gluck whose first name is Christoph Willibald, from Upper Palatinate.’

(455)

„Vermutlich wird der 16-Jährige der Täter sein”, sagt presumably will the 16.year.old the culprit be-inf says Polizeisprecher Thomas Buchheit, der aber nicht ausschließen police.spokesman Thomas Buchheit who yet neg exclude will, dass der junge Mann in Notwehr gehandelt haben wants that the young man in self-defence act-ppp have-inf könnte.³⁷⁶ could ‘ „Presumably, the 16 year old will be the culprit” says the police spokesman Thomas Buchheit, who does not want to exclude that the young man has acted in selfdefence.’

(456)

Es wird schon einen Grund gehabt haben, warum die it will already a reason have-ppp have why the Eisbärin ihre Jungen gefressen hat.³⁷⁷ ice.bear.lady her offspring eaten has ‘There will be a reason why the polar bear has eaten her offspring.’

(457)

So knapp 1000 Zuschauer werden es wohl gewesen sein, die so barely 1000 spectators will it wohl be-ppp be-inf that trotzdem kamen.³⁷⁸ nevertheless came ‘There will have been barely 1000 spectators that came nevertheless.’

(458)

In meinem zwölften Lebensjahr wird es gewesen sein, als am in my twelfth live.year will it be-ppp be-inf as at.the Frühmorgen des heiligen Christ abends mein Vater early.morning the-gen holy-gen christ evening-gen my father mich an der Schulter rüttelte³⁷⁹ me on the shoulder shook ‘It must have been in when I was twelve that my father shook me on my shoulder.’

Furthermore, in all of the instances above the embedded proposition is not part of the speaker’s knowledge. In a similar manner, Comrie (1989: 60), FabriciusHansen (2000: 183–6), Erb (2001: 161) and Krämer (2005: 60, 133) argue that the

375 376 377 378 379

DeReKo: O98/AUG.75935 Neue Kronen-Zeitung, 04/08/1998. DeReKo: HAZ08/MAI.03238 Hannoversche Allgemeine, 19/05/2008. DeReKo:NUN08/JAN.00722 Nürnberger Nachrichten, 08/01/2008. DeReKo: NUN07/APR.00752 Nürnberger Nachrichten, 10/04/2007. DeReKo: K00/DEZ.81948 Kleine Zeitung, 24/12/2000.

228 | 2 Modal verbs: A class struggle

speaker does not have the direct evidence which is necessary to assert the embedded proposition. It is remarkable that most of the instances found in the DeReKo corpus have been taken from interviews, direct speech and other instantiations of spoken language. Furthermore, epistemic werden is frequently used for the estimations of numbers, as is exemplified in (457). There are further parallels between werden and the traditional modal verbs: As Erb (2001: 146) illustrates, epistemic werden can hardly be found as an infinitive or as a past participle. But then, Krämer (2005: 34) demonstrates that epistemic werden is excluded from a couple of environments in which epistemic müssen and können can occur, such as in the scope of negation, in questions, and under verum focus. This leads Krämer (2005: 49) to the conclusion that werden is an extra-propositional evidential operator in opposition to epistemic modal verbs. As this matter is far from trivial, it will be postponed to Section 4. As pointed out above, it is far from obvious which modal force the epistemic use of werden could carry. While Brennan (1993: 97) and Enç (1996: 356) analyse its English counterpart will as a necessity modal, the situation in German is less obvious. Kratzer (1981: 58) appears to be inclined to analyse werden as a necessity modal operator, associating it with the adverb certainly. Zifonun (1997: 1910) concludes that epistemic werden acts as a modal probability operator. In a similar vein, Fritz (1997: 94) concludes that epistemic werden expresses a stronger modal force than epistemic können but a weaker modal force than epistemic müssen. Finally, Vater (1975: 113) localises the modal force of werden somewhere between that of können and müssen. If epistemic werden was indeed a canonical necessity modal operator such as müssen, it should be possible to substitute it with müssen without affecting the overall interpretation. However, such a replacement would change the meaning of the pattern (cf. 456). Moreover, it is fairly questionable that the semantic difference between epistemic müssen and epistemic werden is only a matter of modal force. In order to account for the different interpretations of epistemic must and epistemic will in English, Sweetser (1990: 55) and Ziegeler (2006: 88) conclude that the specific contribution of epistemic will concerns the verification of the embedded proposition. Whereas the future interpretation of will expresses that the speaker expects a certain event to occur in the future, the epistemic reading indicates that the speaker holds the expectation that the speculative embedded proposition can be verified in the future. As Sweetser (1990: 55) concludes, the use of epistemic will is based on the assumption “if we check we will find out”. FabriciusHansen (1999: 124) and Krämer (2005: 133) suggest an analogous analysis for German werden.

2.2 Case studies |

229

Though very plausible, the approaches sketched above faces serious problems. There are uses of epistemic werden in which the speaker definitely knows that he will never be in a position to verify the speculative embedded proposition: (459)

Sie werden es also gewusst haben: Diesmal würden sie they will it thus know-ppp have-inf this.time would they nicht mehr wegkommen.³⁸⁰ neg more away.get-inf ‘They will/must have known that they would not be to late to escape this time.’

Example (459) is taken from a text that deals with the suicide of the national socialist terrorists Uwe Bönhardt and Uwe Mundlos. In their final moment, their bus was surrounded by the police, there was no way of escaping. Under unknown circumstances the two terrorists shot themselves. Using the epistemic instance of werden, the author of that article attributes a thought to the two men which they could have had in mind leading up to their suicide. It is fairly doubtful if the author really is convinced that he will find out the last thoughts of these terrorists. As it turns out, the suggestion made by Sweetser (1990: 55) is not entirely convincing and requires slight modifications.

2.2.11 scheinen and dünken While werden and brauchen exhibit patterns that are entirely parallel to the epistemic readings of modals, the raising verb scheinen ‘seem, appear’ evidently differs in a couple of respects. First of all, it is restricted to zu-infinitive complements. In opposition to brauchen, no instances with a bare infinitive complement have been observed so far. Moreover, it always exhibits some sort of epistemic or evidential interpretation, except for its unergative use. Finally, scheinen optionally realises the epistemic attitude holder as dative NP. There is a lively debate going on to what extent scheinen with zu-infinitive can be subsumed under the same syntactic and semantic class as epistemic modal verbs. On the one hand, some authors, such as Askedal (1998: 61) and Wurmbrand (2001: 205), argue that scheinen is an epistemic verb, just like the traditional epistemic modal verbs. On the other hand, Pafel (1989: 143) holds that scheinen behaves differently in a couple of essential points: It can optionally realise the epistemic attitude holder as its indirect object NP. As has been noticed by Ebert (1976: 41–45) and Diewald (2001: 101), scheinen is similar to dünken ‘seem, to cause to think’ in many respects, which used to be

380 Der Spiegel, 47/2011 21.11., p.22.

230 | 2 Modal verbs: A class struggle

the causative counterpart of denken ‘think’. However, the latter element has almost disappeared over the course of the last centuries. As dünken behaves slightly differently, it will be useful to determine the nature of scheinen. The verb scheinen occurs in five syntactic patterns: (i) as an unergative intransitive verb, (ii) as a copula, (iii) as an impersonal verb with a finite dass-clause complement, (iv) as an impersonal verb that selects a hypothetical comparative als ob-clause, and (v) as a raising verb that selects a zu-infinitive complement. Similar classifications have been proposed by Pafel (1989: 124), Askedal (1998: 70), Diewald (2001: 94) and Lima (2004). In opposition to that, dünken can only be found in the configurations (ii)–(v) depending on the respective historical period. In Contemporary Standard German, dünken has almost disappeared, though is is still occasionally used as a copula. 2.2.11.1 Unergative uses As has been illustrated by Diewald (2001: 95) and Lima (2004), scheinen can still be used as an intransitive main verb with a clear lexical meaning. In this use, it determines its subject referent as a source that emits light, as is exemplified in (460)–(461). (460)

Der fast runde Mond scheint am Himmel.³⁸¹ the almost round moon shines at.the sky ‘The almost round moon is shining in the sky.’

(461)

Wie gut ist es, wenn in der Dunkelheit Lichter scheinen und das How good is it when in the darkness light shine and the Dunkle hell machen.³⁸² dark bright make ‘How good it is when there are lights shining in the darkness and they make the dark become bright.’

Occasionally, unergative scheinen is complemented by a directional phrase that describes the path the ray of light follows. 2.2.11.2 Predicative phrases As has been illustrated by Askedal (1998: 52–53) and Diewald (2001: 95) and Lima (2004) scheinen can be used as a copula that selects various types of predicative phrases: adjectives (cf. 462), noun phrases (cf. 463) and prepositional phrases. In

381 DeReKo: A09/NOV.00303 St. Galler Tagblatt, 02/11/2009. 382 DeReKo: M11/DEZ.03530 Mannheimer Morgen, 10/12/2011.

2.2 Case studies |

231

its copula pattern, scheinen is similar to English seem in that it expresses that somebody has the evidence to conclude that the embedded predication is valid. In opposition to other copulas such as sein, werden or bleiben ‘stay’, scheinen optionally realises a dative argument that refers to the epistemic attitude holder who is exposed to the unspecified type of evidence. As Askedal (1998: 52–53) points out, scheinen is found with all types of predicative phrases, with as well as without a dative object. The corpus study presented here has revealed three interesting tendencies. First of all, the copula scheinen does not occur very frequently with genuine predicative NPs, as in (463). Furthermore, the dative object of scheinen is not very often realised. Finally, if it is realised, it is almost always represented by a 1st person pronoun (mir or uns). Examples with other types of dative NPs are fairly rare, though they do occur (cf. 463): (462)

Doch ihr Ehrgeiz scheint ungebrochen.³⁸³ yet her ambition seems unbowed. ‘Yet, her ambition seems unbowed.’

(463)

Gewaltloser Widerstand scheint ihm das einzig probate non-violent resistance appears him-dat the only appropriate Mittel.³⁸⁴ means ‘Non-violent resistance appears to be the only appropriate means to him.’

The verb dünken behaves like scheinen in many respects. It can be combined with predicative adjectives and predicative nouns. Its object is almost always specified for the 1st person. However, it differs in two respects. Firstly, dünken realises its epistemic attitude holder as an accusative NP, rather than a dative NP. Secondly, this accusative NP appears to be obligatory and is present in most of the instances found in the DEREKO corpus. (464)

Der Weg nach Tobel dünkt ihn unendlich weit.³⁸⁵ the way to Tobel thin-caus him-acc eternally long ‘The way to Tobel seems eternally long.’

(465)

Eine Ewigkeit dünkt mich das Warten.³⁸⁶ a eternity think-caus me-acc the waiting ‘The waiting seems an eternity to him.’

383 384 385 386

DeReKo: M06/JAN.01737 Mannheimer Morgen, 09/01/2006. DeReKo: HAZ08/NOV.04416 Hannoversche Allgemeine, 25/11/2008. DeReKo: A98/FEB.12719 St. Galler Tagblatt, 28/02/1998. DeReKo: O95/MAI.44351 Neue Kronen-Zeitung, 05/05/1995.

232 | 2 Modal verbs: A class struggle

As dünken is an archaic pattern, it becomes less and less frequent. Most contemporary native speakers contaminate the fairly particular pattern of dünken with semantically related patterns of verbs that occur more frequently, such as scheinen. Accordingly, the accusative NP alternates with a dative NP, as will be exemplified in the upcoming sections. Moreover, instances can be found in which dünken has a regular preterite form. 2.2.11.3 dass-clauses Pafel (1989: 124), Askedal (1998: 52) and Diewald (2001: 98) have illustrated in great detail that scheinen can be used as an impersonal verb that subcategorises a finite dass-clause. The interpretation is analogous to the copula pattern of scheinen. As Askedal (1998: 52) has observed, the realisation of the dative object is optional. Once again, this dative object is mostly instantiated as a 1st person pronoun. In some rare cases, other types of NPs can be found in this context as well, as is exemplified in (466). As regards dünken, it behaves nearly analogously to scheinen, except for the fact that it has an obligatory object. As example (469) indicates, some speakers tend to use a dative NP rather than an accusative one, possibly due to contamination with the scheinen-pattern. (466)

Ihm scheint, dass die Zeit der touristischen him-dat seems that the time the-gen tourist Alpen mit grosstechnischen Anlagen Erschliessungen der sites development the-gen alps-gen with big.technical vorbei ist.³⁸⁷ over is ‘It seems to him that the time of tourist development in the alps with colossal technical sites is over.’

(467)

Es scheint, dass er sich an das Leben in freier Natur gewöhnt it seems that he refl on the live in free nature accustomed hat³⁸⁸ has ‘It seems that it has become accustomed to the life in the great outdoors.’

(468)

Aber mich dünkt, dass der Weg dorthin immer länger but me-acc think-caus that the way there always long wird.³⁸⁹ becomes

387 DeReKo: A01/DEZ.48633 St. Galler Tagblatt, 03/12/2001. 388 DeReKo: A08/APR.03463 St. Galler Tagblatt, 09/04/2008.

2.2 Case studies |

233

‘Yet, it appears to me that it the way there becomes longer and longer.’

(469)

Mir dünkt, dass Bauvorhaben der öffentlichen me-dat thinks-caus that building.projects the-gen public-gen Hand mit Absicht so niedrig wie möglich kalkuliert hand-gen with intention as low as possible calculate-ppp werden, damit sie beim Steuerzahler als günstig aux.pass so.that they at.the tax.payer as cheap erscheinen.³⁹⁰ appear ‘It seems to me that building projects of the public authorities are intentionally calculated to be as cheap as possible in order to make them look like a bargain to the tax payer.’

As Askedal (1998: 52) has pointed out, the complement clause can be alternatively realised as a clause that exhibits a verb second word order, and that does not include any subordinating conjunction. As this alternation does not affect only scheinen with finite dass-clause but a very extensive class of verbs, it will not receive any further attention here. As Diewald (2001: 104) emphasises, scheinen with finite dass-clause occurs from the 18th century onwards, whereas the pattern with zu-infinitive complements already emerged in the 16th century, as has been illustrated by Ebert (1976: 41). Accordingly, the pattern with zu-infinitive cannot be derived from dassclauses, as is sometimes suggested, for instance by Chomsky (1981: 43). Moreover, Ebert (1976: 41) has demonstrated that dünken with daz-clause is already documented in the Middle High German pârzival. 2.2.11.4 Hypothetical comparative als ob-clauses Askedal (1998: 53) points out that scheinen is moreover found with finite als obclauses and als-clauses. These types of subordinate clauses strikingly resemble hypothetical comparative clauses, as they are characterised in Jäger (2010: 469). In a similar way, dünken can be combined with hypothetical comparative clauses as well. As far as argument structure is concerned, both verbs behave in the same way as they do with dass-clause complements: In the case of scheinen the realisation of the dative NP is optional (cf. 470–471), in the case of dünken the realisation of the argument is obligatory; once again, accusative and dative alternate (cf. 472–473).

389 DeReKo: A11/JUL.02254 St. Galler Tagblatt, 07/07/2011. 390 DeReKo: RHZ11/JUL.06178 Rhein-Zeitung, 06/07/2011.

234 | 2 Modal verbs: A class struggle

(470)

es scheint mir, als ob du heute nicht irren könntest.³⁹¹ it seems me-dat as if you today neg err could ‘It seems to me as if you could not err today.’

(471)

Es scheint nur so, als ob Wolfgang Schäuble nun gerade it seems only so as if Wolfgang Schäuble now just deswegen neue Gesprächsbereitschaft signalisiert: Öffentlich, therefore new talk.readiness signals in.public über die Medien, wandte er sich schon vor der von den through the media addressed he refl already before the by the Währungshütern verabreichten Ohrfeige an die currency.watchdog administered slap to the Sozialdemokraten.³⁹² Social.democrats ‘It only seems as if Wolfgang Schäuble signals new readiness to negotiate for that reason. Yet, by media he had already addressed to the Social Democrats before they had got their slap by the currency watchdog.’

(472)

Eine Nacht hatte er auf der Harzburg einen schweren Traum; es a night had he at the Harzburg a heavy dream it deuchte ihm, als ob er mit einem furchtbaren Eber think-pst.caus him-dat as if he with a terrible boar kämpfe, der ihn nach langem Streit zuletzt besiegte.³⁹³ fight-sbjv.prs that him after long fight finally overwhelmed ‘One night at the Harzburg, he had an oppressive dream. It seemed to him as if he were fighting with a terrible boar that overwhelmed him after a long battle.’

(473)

Es dünkt mich, als ob alle ein Lächeln auf den Lippen it thinks-caus me-acc as if all a smile on the lips hätten³⁹⁴ had ‘It seems to me as if everybody had a smile on his lips.’

As the instance in example (472) indicates, the tendency to replace the accusative NP with a dative NP is not a recent development. It is already found in the collection of tales edited by the Grimm Brothers in the early 19th century. Example (471) reveals the true nature of the verb. The speaker clearly distinguishes between the mere appearance of a state of affairs and its factual being. He

391 DeReKo: GOE/AGM.07859 Goethe: Wilhelm Meisters Wanderjahre, [Roman], (Erstv. 1821). 392 DeReKo: NUN97/AUG.01105 Nürnberger Nachrichten, 14/08/1997. 393 DeReKo: GRI/SAG.00311 Des Hackelnberg Traum, (Erstv. 1816 ; 1818), In: Deutsche Sagen, gesammelt von Jacob und Wilhelm Grimm. – o.O., 1891. 394 DeReKo: SOZ11/DEZ.05459 Die Südostschweiz, 27/12/2011.

2.2 Case studies |

235

is totally aware that the willingness to negotiate signalled by Schäuble is not new. He just wants to stress that it only looks as if it were new. Using scheinen, it is possible for the speaker to refer to the mere appearance of a state of affairs, even if he knows that it is false. A similar example has been provided by Colomo (2011: 225). Canonical epistemic modal verbs cannot be used in this way unless they are in the scope of a counterfactual operator, as has been shown in Section 2.2.6, 2.2.1 and 2.2.2. But in such configurations it remains to be checked what influence the counterfactual operator has. Finally, scheinen in example (471) appears to carry verum focus. If the sentence was uttered, it is very likely that the verb scheint would bear focal stress. The alternatives under discussion are {appearance, factual being}. In the specific example given above, the focus particle nur would clearly refer to scheint, blocking all of the remaining alternatives. Interestingly, the effect is reminiscent of verum focus, as has been discussed by Höhle (1992). A similar effect can be observed with some epistemic modal verbs, as is illustrated in Section 4.9 and 4.10. These facts reveal the true nature of the verbs under analysis: They compare two state of affairs. In the context of example (471), the speaker faces a state of affairs which exhibits the characteristics as in the hypothetical state of affairs in which Wolfgang Schäuble signals willingness to negotiate. The first state of affairs is syntactically represented by the subject es, and the second state of affairs is realised as the hypothetical comparative als ob-clause. As a consequence, the semantic contribution of scheinen is to introduce a hypothetical state of affairs, to which the actual situation is compared. This also reflects the nature of its original use as a copula, in which two properties are associated to each other. 2.2.11.5 Raising infinitives with clause modification Finally, it has been demonstrated by Ebert (1976: 41), Pafel (1989: 124), Askedal (1998: 52) and Diewald (2001: 97) that scheinen selects a zu-infinitive complement. There is a widespread consensus that scheinen does not select a referential subject and that, consequently, it has to be a raising verb. Once again, the realisation of the dative object is optional in the case of scheinen, as is illustrated in examples (474)–(477). However, as Askedal (1998: 56) has found out, the dative argument occurs far less frequently whenever scheinen selects a zu-infinitive complement, specifically in only 5.22 % of the cases of his corpus. By contrast, scheinen with finite clausal complements realises the dative NP in 38,52 % of the cases, and scheinen with a predicative complement in even 56,05 % of the cases. Nevertheless, it can be found in the DeReKO corpus, as illustrated in examples (475)–(477). Another example is provided by Askedal (1998: 52).

236 | 2 Modal verbs: A class struggle

In opposition to scheinen, the use of dünken with zu-infinitive complement has almost disappeared from the contemporary language, as has been shown by Maché and Abraham (2011: 266). In earlier stages, it was more frequent in this configuration, and it even exhibited a IPP morphology in the 16th century (cf. 479). (474)

Jospin scheint das Opfer seines eigenen Erfolgs Jospin seems the victim his-gen own-gen success-gen geworden zu sein.³⁹⁵ become-ppp to be-inf ‘Jospin seems to have become the victim of his own success.’

(475)

Eher das Gegenteil scheint mir der Fall zu sein.³⁹⁶ rather the opposite seems me-dat the case to be-inf ‘ It seems to me that rather the opposite is the case.’

(476)

Jede der Figuren scheint mir in ihren Kokon each the-gen character-gen seem me-dat in her cocoon eingesponnen zu sein.³⁹⁷ form-inf to be-inf ‘It seem to me that each character is caught in his own cocoon.’

(477)

Deren Gesichter schienen ihm wie in einem rot gepunkteten their faces seem him-dat like in a red dotted Nebel zu verschwimmen.³⁹⁸ mist to blur-inf ‘It seemed to him as if their faces blurred in a red dotted mist.’

(478)

Lohnverzicht und Abkehr vom 30-Stunden-Modell, das wage.sacrifice and renunciation from.the 30.hours.model that dünkt sie denn doch reines Teufelswerk zu sein.³⁹⁹ think-sbjv.pst them-acc then part pure devil.work to be-inf ‘It seems to them that the sacrifice of their wage and the renunciation from the 30 hours model is the pure work of the devil.’

(479)

395 396 397 398 399

Vnd sagte jm / wie er die Portugaleser hette and said him how he the Portuguese have-sbjv.pst lernen kennen / vnd er were allwegen jr learn-ppp(ipp) know-inf and he be-sbjv.pst always their

DeReKo: E00/MAR.07351 Zürcher Tagesanzeiger, 17/03/2000. DeReKo: SOZ10/MAI.04445 Die Südostschweiz, 25/05/2010. DeReKo: BRZ07/JAN.19721 Braunschweiger Zeitung, 25/01/2007. DeReKo: BRZ06/MAR.07145 Braunschweiger Zeitung, 13/03/2006. DeReKo: NUN95/NOV.00207 Nürnberger Nachrichten, 03/11/1995.

2.2 Case studies |

237

Freund gewesen / denn sie hetten jn friend be-ppp since they have-sbjv.pst him-acc rechtschaffene Leut duencken seyn /⁴⁰⁰ righteous people think-caus.ppp(ipp) be-inf ‘And he said to him how he made acquaintance with the Portuguese and that he always was their friend as they seemed to him righteous people.’

As examples (480) and (481) illustrate, scheinen allows for the selection of infinitives with non-referential subject NPs. This indicates that it has to be analysed as a raising verb indeed. (480)

Unter den Bier-Fans scheint es wesentlich mehr Kampftrinker among the beer-fans seem it considerably more binge.drinker zu geben als unter den Wein-Freunden.⁴⁰¹ to give-inf than among the wine-friends ‘Among the fans of beer, there seem to be considerably more binge drinkers than among the friends of wine.’

(481)

In Deutschland scheint es elf Monate im Jahr zu regnen.⁴⁰² in Germany seem it eleven months in.the year to rain-inf ‘In Germany, it seems to rain eleven months a year.’

As has been mentioned above, some authors group scheinen with zu-infinitive together with the traditional epistemic modal verbs into a single syntactic and semantic class. Askedal (1998: 60) illustrates that scheinen involves (i) a raising pattern, (ii) it selects an obligatorily coherent infinitive complement/it triggers clause union, (iii) it lacks an imperative, (iv) it lacks an infinitive, (v) it lacks a past participle, and (vi) it cannot select VP-anaphora, as has also been illustrated by Neugeborn (1976). These observations lead Askedal (1998: 61) to the conclusion that scheinen with a zu-infinitive complement belongs to the class of future and epistemic auxiliaries, together with the traditional epistemic modal verbs werden, drohen, versprechen and pflegen. For similar reason, Wurmbrand (2001: 205) argues that scheinen belongs to the class of epistemic (modal) verbs. In any case, it cannot be denied that there are analogies, the question is how essential they are. However, there are some respects in which scheinen behaves fairly differently from the traditional epistemic modal verbs. Firstly, (i) scheinen grammatically realises the epistemic attitude holder, as has been pointed out by Pafel (1989: 125). Apart from that, (ii) it behaves in a peculiar way when it bears past morphology.

400 Ulrich Schmid Neuwe Welt, p. 20b, (1567). 401 DeReKo: RHZ96/AUG.07223 Rhein-Zeitung, 14/08/1996. 402 DeReKo: NUZ06/JUN.00564 Nürnberger Zeitung, 06/06/2006.

238 | 2 Modal verbs: A class struggle

In this type of context, it refers to past evidence. Nevertheless, the Time of Evaluation can be identical to the Time of Utterance, as will be illustrated in more detail in Section 5.2. In these environments, it cannot be replaced with an epistemic modal with past morphology. Moreover, (iii) scheinen introduces an infinitive particle zu that cannot be dropped under any circumstance. Furthermore, (iv) the speaker can know that the embedded proposition is in fact false, as has been demonstrated by Colomo (2011: 225). This is a fairly unlikely scenario with epistemic modal verbs and obviously only possible if they are in the scope of a counterfactual operator, as has been demonstrated in Sections 2.2.1, 2.2.2 and 2.2.6. Finally, (v) scheinen does not involve any interpretation that could be considered as a circumstantial modality, or future oriented reading as it is possible with werden. All of these observations indicate that scheinen is a predicate that behaves in a similar way with respect to epistemic modal verbs. Yet, it turns out that it is substantially different from epistemic modal verbs, and it occurs in many contexts where epistemic modal verbs are totally excluded. Apart from that, any analysis which considers scheinen as an epistemic modal verb neglects the fact that it also embeds hypothetical comparative clauses. Accordingly, scheinen is a verb which compares two state of affairs. Even if it is less explicit, this analysis also applies to scheinen with an infinitive complement. Uttering example (474), the speaker expresses that he perceives a state of affairs in the real world which exhibits a subset of essential characteristics which are identical with the hypothetical state of affairs in which Jospin is the victim of his own success. As with the CoDeC for epistemic modal verbs, the speaker indicates with respect to the proposition {Jospin is the victim of his own succes} that he does not perceive all the essential characteristics which are identical with the hypothetical state of affairs in which Jospin is the victim of his own succes. This could be a result of Grice’s (1989: 26) Maxim of Quantity Q2, which states that a speaker should not make his contribution more informative than is required.

2.2.12 drohen, versprechen and verheißen As has been illustrated by Bech (1955: 126–127), Kiss (1995: 154), Askedal (1997b), Reis (2001: 312 Fn. 40, 2005, 2007), Wurmbrand (2001: 205), Lima (2005) and Colomo (2011), drohen ‘threaten’ and versprechen ‘promise’ can be used as raising verbs that appear to be related to epistemic modal verbs. Moreover, Łukasz Jędrzejowski (pers. commun.) has suggested that verheißen ‘augur, promise’ is another promising candidate that could occur in this pattern as well. And indeed, it can be found in a very similar distribution as versprechen.

2.2 Case studies |

239

This phenomenon is not restricted to German but it can be found in several European languages such as English promise and threaten (cf. Traugott (1997)), Portuguese ameaçar ‘threaten’ (cf. Lima (2005)) and Spanish prometer ‘promise’ and amenazar (cf. Cornillie (2007: 85)). Once again, there are accounts in which drohen and versprechen are explicitly analysed as epistemic modal verbs, e.g. Askedal (1997b: 14), Wurmbrand (2001). In contrast to that, there are also approaches in which the opposite is assumed. Reis (2001: 312 Fn. 40, 2005: 129, 2007: 13) and Colomo (2011) argue that drohen and versprechen differ from epistemic modal verbs in essential respects. Accordingly, they should not be treated as epistemic modal verbs. As will be seen in the empirical investigations presented below, the latter approach has more explanatory potential. As will be shown, all of the verbs under discussion occur in various syntactic patterns. However, they do not behave in a uniform way. First of all, drohen is employed (i) as a two place intransitive verb that selects a dative NP and a mitPP (‘with’-PP); (ii) the referent expressed by the mit-PP can be realised as a finite dass-clause; and finally, (iii) drohen occurs with zu-infinitive complements. In this configuration, it can either involve a (iii) control pattern, or a (iv) raising pattern. By contrast, versprechen and verheißen occur as (i) ditransitive verbs with an accusative NP and a dative NP, (ii) with finite dass-clause complements, and, finally, with control zu-infinitive complements and raising zu-infinitive complements. 2.2.12.1 Transitive and intransitive uses All of the three verbs under discussion occur as three-place predicates that select an agent argument, a recipient argument and a theme (or perhaps instrument) argument. In any case, the agent argument is realised as a nominative NP and the recipient argument as a dative NP. The remaining argument is represented as a mit-PP rather than as an accusative NP in the case of drohen. In (482), this argument surfaces as the PP mit Krieg ‘with war’. Accordingly, this particular use of drohen cannot be considered as a transitive use. By contrast, the third argument is realised as an accusative NP in the case of versprechen and verheißen. In the examples below, versprechen realises its theme argument as the accusative NP anstrengungslosen Wohlstand ‘effortless wealth’ (cf. 483), and verheißen as the accusative NP kein grosses Outperformance-Potential (cf. 484). Likewise, Colomo (2011: 221) observes that drohen and versprechen differ in terms of argument structure.

240 | 2 Modal verbs: A class struggle

(482)

Kabila drohte unterdessen dem Nachbarland Ruanda Kabila threatened meanwhile the neighbour.country ruanda mit Krieg, weil es an der Seite der Rebellen in den with war because it at the side the-gen rebels-gen in the Kongo einmarschiere.⁴⁰³ Kongo invade-sbjv.prs ‘Kabila threatened his neighbour country ruanda with war as it supposedly invades the Kongo together with the rebels.’

(483)

Wer dem Volk anstrengungslosen Wohlstand verspricht, whoever the people effortless wealth promises lädt zu spätrömischer Dekadenz ein.⁴⁰⁴ invites to Late.Roman decadence in ‘Those who promise effortless wealth to the people will yield Late Roman decadence.’

(484)

Der CSFB-Analyst verheisst der Schweizer Börse the CSFB-analyst augurs the Swiss Stock.Exchange deshalb „kein grosses Outperformance-Potential”.⁴⁰⁵ therefore no big out.performance.potential ‘The CSFB-analyst predicts no big out-performance-potential for the Swiss Stock Exchange.’

As it turns out, the dative NP is not realised in most of the instances found in the corpus. In this respect, examples (482)–(484) illustrated above are rather atypical. 2.2.12.2 dass-clauses Drohen, versprechen and verheißen occur as three-place predicates which realise their theme argument as a finite dass-clause. This merits closer attention, as the three verbs differ in other environments with respect to the realisation of the theme argument. However, whenever the theme argument surfaces as a clausal argument, these differences disappear. As the instances (486) and (487) indicate, the recipient argument can once again be represented as a dative NP. (485)

Er droht aber auch, dass er Ali Sagdas das Geschlechtsteil he threatens but also that he Ali Sagdas the sex abschneiden und es ihm in den Mund legen werde.⁴⁰⁶ cut.off-inf and it him in the mouth put-inf will-sbjv.prs

403 DeReKo: NUN98/AUG.00548 Nürnberger Nachrichten, 07/08/1998. 404 DeReKo: BRZ10/FEB.08056 Braunschweiger Zeitung, 16/02/2010. 405 DeReKo: E98/MAR.07874 Zürcher Tagesanzeiger, 21/03/1998.

2.2 Case studies |

241

‘But he also threatens Ali Sagats to cut off his sex and to put it into his mouth.’

(486)

Generell versprechen Union und FDP den Firmen, dass es in generally promise Union and FDP the enterprises that it in Zukunft unbürokratischer zugeht.⁴⁰⁷ future non.bureaucratic goes ‘Generally, the Union and the FDP promise to the enterprises that the procedure will be less bureaucratic in future.’

(487)

Seinem ehemaligen Salzburger Assistenten Philippe Auguin his former Salzburger assistant Philippe Auguin verheißt er, daß er in Nürnberg „ganz sicher ein promises he that he in Nürnberg very certainly a ,Meisterdirigent‘ wird!”⁴⁰⁸ master.conductor becomes ‘He promises to his former assistant in Salzburg Philippe Auguin that he will become a master conductor in Nürnberg.’

Once again, the dative argument occurs only fairly rarely, irrespective of the matrix predicate. In most of the instances that can be found in the corpus, the recipient is not overtly realised. 2.2.12.3 Control infinitives with event modification Alternatively, drohen, versprechen and verheißen occur as three-place predicates, which realise their theme arguments as zu-infinitive complements. As Reis (2005: 126, 2007: 8) and Colomo (2011: 142) indicate, these instances have to be considered as control verbs, since they impose selectional restrictions on their subject arguments. In their terms, these verbs are commissive speech act verbs that involve a subject referent who makes a commitment concerning the state of affairs expressed by the infinitive complement. In all of the examples (488)–(490), the matrix subject referent intentionally commits the communicative act indicated by the matrix predicate. Accordingly, the subject NP must be an argument of the respective predicate. (488)

Er droht ihr, sie in ein Heim zu stecken.⁴⁰⁹ he threatens her her into a asylum to put-inf ‘He threatens (her) to put her into a children’s home.’

406 407 408 409

DeReKo: NON09/DEZ.14663 Niederösterreichische Nachrichten, 28/12/2009. DeReKo: RHZ09/OKT.14552 Rhein-Zeitung, 17/10/2009. DeReKo: NUN98/FEB.00803 Nürnberger Nachrichten, 10/02/1998. DeReKo: BRZ09/JUL.22822 Braunschweiger Zeitung, 21/07/2009.

242 | 2 Modal verbs: A class struggle

(489)

Als seine Mutter im Sterben liegt, verspricht er ihr widerwillig, as his mother in dying lies promises he her begrudgingly den Jakobsweg zu gehen.⁴¹⁰ the Jakob.Way to go-inf ‘ When his mother was about to die, he promised her to do a pilgrimage along the way of St. James.’

(490)

Als Gott in Gestalt dreier Männer Abraham und Sara besucht as God in guise three-gen men-gen Abraham and Sara visits und der alternden Frau verheißt, einen Sohn auf die Welt zu and the ageing woman augurs a son on the world to bringen, bricht sie unwillkürlich in Lachen aus (Gen 18).⁴¹¹ put-inf breaks she involuntarily in laughs out Gen 18 ‘When God visited Abraham and Sara in guise of three men and augured to the aging woman that she will give birth to a son she involuntarily burst out laughing (Gen 18).’

Once again, it is possible to realise the recipient argument as a dative NP. However, such cases do not occur frequently in the DeReKo corpus. Further occurrences from corpora are provided by Müller (2002: 55-6) 2.2.12.4 Raising infinitives with event modification Finally, drohen, versprechen and verheißen occur in configurations in which they exhibit a fairly different interpretation. While all of the uses discussed above belong to the class of commissive speech act verbs, they sometimes appear as uses in which their precise meaning cannot be captured easily. Thus, some authors conclude that they are modal or aspectual auxiliaries. Accordingly, Drosdowski et al. (1984: 94) argue that drohen acts as a modifizierendes Verb, i.e. as a modifying verb. Yet, for some reason they do not make any mention of an analogous use of versprechen. In an earlier edition of the Duden-grammar, Grebe et al. (1966: 528), both verbs are documented with a interpretation different from their commissive speech act reading. As is commonly assumed, this difference in interpretation is reflected in a drastic change in the argument structure as well. Based on the observation made by Bech (1955: 113, 126), most authors, such as Askedal (1997b: 13), Reis (2005: 127, 135, 2007: 21, 32), Wurmbrand (2001: 205), conclude that these uses lack a referential subject argument and the recipient argument. As Colomo (2011: 233) points out, the raising uses of drohen and versprechen behave very

410 DeReKo: RHZ07/MAI.13875 Rhein-Zeitung, 14/05/2007. 411 DeReKo: RHZ07/FEB.09600 Rhein-Zeitung, 10/02/2007.

2.2 Case studies |

243

differently from the raising verb scheinen in this respect, which can realise its dative object under any circumstance. As examples (491) and (493) indicate, these verbs can select infinitives that do not involve referential subject arguments. Similar examples are provided by Müller (2002: 55-6) Reis (2005: 135, 2007: 21). Moreover, Reis (2005: 139, 2007: 27) demonstrate that drohen is transparent with respect to voice: It will yield the same interpretation no matter whether it embeds a proposition based on an active verb or its passivised counterpart. All of these facts indicate that drohen and versprechen involve a variant that has to be analysed as a raising pattern. (491)

Es droht zu regnen, doch das stört die vielen Schwimmer it threatens to rain-inf but this bothers the many swimmers an diesem Sonntagmorgen kaum.⁴¹² on this Sunday.morning hardly ‘It threatens to rain but this does not bother the many swimmers on this Sunday morning.’

(492)

Selten war die Kampfkraft der „Bild”-Zeitung so rarely was the fighting.power the-gen Bild-newspaper so beansprucht wie in der vergangenen Woche, als Doktor challenged as in the passed week when Doktor Guttenberg zerschossen zu werden drohte.⁴¹³ Guttenberg obliterate-ppp to become-inf threatened ‘The fighting power of the „Bild” newspaper was rarely so challenged as in the past week when Doktor Guttenberg was in threat of being dishonoured.’

(493)

Es verspricht, ein schöner Tag zu werden.⁴¹⁴ it promises a beautiful day to become-inf ‘It promises to be a beautiful day.’

(494)

„Juchzet, frohlocket!” heißt es ganz programmatisch, und das cheer rejoice calls it very programmatically and the dargebotene Liedgut verspricht so vielfältig zu sein wie presented repertoire promises so manifold to be-inf as des Showmans Kostüme.⁴¹⁵ the-gen showman-gen costume ‘ “Cheer, rejoice!” it says very programmatically and the presented repertoire promises to be as manifold as the showman’s costumes.’

412 413 414 415

DeReKo: NUZ09/JUN.02681 Nürnberger Zeitung, 29/06/2009. Spiegel 9/2011, 28.2. 2011, p. 141. DeReKo: HAZ08/MAI.05336 Hannoversche Allgemeine, 29/05/2008. DeReKo: HMP06/DEZ.02074 Hamburger Morgenpost, 21/12/2006.

244 | 2 Modal verbs: A class struggle

As Łukasz Jędrzejowski (pers. commun.) has suggested, verheißen, due to its semantic affinity to versprechen, is another promising candidate for the raising pattern sketched above. And indeed, it is occasionally found in raising-like configurations (cf. 495–496). (495)

Ein Höhepunkt verheißt das Wochenende vom 22. bis zum a highlight promisses the weekend from 24 until to.the 24. Mai zu werden.⁴¹⁶ 24 may to become-inf ‘The weekend from 22nd until 24th of May promises to become a highlight.’

(496)

Mit den sechs neuen Vereinen, den beiden Viernheimer Clubs, with the six new clubs the both Viernheimer clubs St. Ilgen, Leimen, Bammental und Treschklingen verheißt es St. Ilgen Leimen Bammental and Treschklingen promisses it eine spannende Runde zu werden.⁴¹⁷ a exciting season to become-inf ‘With the six new clubs, the two clubs from Viernheim, St. Ilgen, Leimen, Bammental and Treschklingen, it promises to be an exciting season.’

Unfortunately, due to the low frequency of occurrences, no instances could be found that unambiguously exemplify the diagnostics of a raising verb. The classification as a raising verb is based on mere semantic analogies to the raising verb versprechen. As has been indicated above, the three verbs do not behave in the same way. First of all, there is a subtle difference in the argument structure concerning the realisation of the theme argument. Secondly, it has been observed by Askedal (1997b: 17) and Diewald and Smirnova (2010: 205, 214) that the raising uses of versprechen are considerably rarer than the raising uses of drohen. According to Askedal, only 12 tokens out of 650 (1.84 %) are raising verbs in the case of versprechen. By contrast, drohen occurs as a raising verb in 96 cases out of 279 (35.7 %). This contrast calls for an explanation. Thirdly, the raising use of versprechen is almost restricted to the selection of the infinitive zu werden, cf. Colomo (2011: 237) for a related observation. Moreover, the three verbs have a couple of characteristics in common. To start with, it has often been remarked that the raising uses of drohen, versprechen and verheißen are restricted to the selection of predicates that refer to an event in the future, cf. Reis (2001: 312 Fn. 40, 2005: 130, 2007: 14) and Colomo (2011: 236).

416 DeReKo: RHZ09/FEB.23166 Rhein-Zeitung, 25/02/2009. 417 DeReKo: M06/AUG.65563 Mannheimer Morgen, 19/08/2006.

2.2 Case studies |

245

In this respect, these verbs crucially differ from the epistemic modal verbs discussed above and from the raising verb scheinen. Moreover, there are semantically related variants of these verbs that do not select an infinitive complement, as Askedal (1997b: 15), Reis (2007: 14) and Diewald and Smirnova (2010: 194, 208) have illustrated. Sometimes, these patterns are referred to as ‘uses with non-agentive/inanimate subject referent’. Regarding this property, drohen, versprechen and verheißen behave very differently from the traditional epistemic modal verbs, which always require an infinitive complement in order to express an epistemic meaning. Again, drohen is by far the most frequent verb in this configuration. By contrast, versprechen and verheißen only occur occasionally in this type of pattern. (497)

Nach einem relativ milden und sonnigen Tag gestern, drohen after a relatively mild and sunny day yesterday threaten dem Urlaubsparadies ab heute neue Unwetter.⁴¹⁸ the holiday.paradise from today new thunderstorms ‘After the relatively mild and sunny day yesterday, the holiday paradise is threatened by new thunderstorms today.’

(498)

5-Tage-Prognose: Das Wochenende verspricht uns kaum Sonne.⁴¹⁹ 5-day-forecast the weekend promises us hardly sun ‘5-day-forecast: The weekend does not promise us much sun.’

(499)

Anderes Wetter verheisst uns der Silvestertag.⁴²⁰ other weather augurs us the new.year.eve ‘The New Years Eve promises us a different weather.’

In contrast to their relatives that involve a raising structure, all of the verbs can be found with a dative object. As the interpretation is almost identical, it is somehow surprising that the recipient argument can be realised as long as no infinitive complement is selected. The dative object is most often found with drohen, whereas it is rarely realised with versprechen and verheißen. As it turns out, the raising uses of drohen, versprechen and verheißen are the result of a rather recent process. The earliest examples of drohen with a raising infinitive complement date from the 18th century. In examples (500)–(501), a commissive interpretation is not very plausible. (500)

O lindre mein Gefühl! – die Brust droht zu zerspringen⁴²¹ o allay my sentiment the breast threatens to burst-inf

418 DeReKo: M01/NOV.86982 Mannheimer Morgen, 14/11/2001. 419 DeReKo: V99/MAR.11078 Vorarlberger Nachrichten, 05/03/1999. 420 DeReKo: A98/DEZ.83932 St. Galler Tagblatt, 28/12/1998.

246 | 2 Modal verbs: A class struggle

‘O, allay my sentiment – the breast threatens to burst.’

(501)

Hochher tobt er in hüpfendem Sprung, und zerschmetterte high riots he in bouncing jump and shattered Waldung // Kracht; doch stets unaufhaltsam enttaumelt er, bis forestry cracks yet always inexorably tumbles he until er erreichet // Ebenen Grund; dann rollt er nicht mehr, wie he reaches even ground then rolls he neg more how gewaltig er andrang: Also droht’ auch Hektor powerfully he closer.gets accordingly threatened also Hektor zuerst, bis zum Ufer des Meeres // Leicht first until to.the shore the-gen sea-gen easy Schiff’ und Gezelte, Danaer hindurchzudringen der pass the-gen Danaian-gen ship and tents-gen Mordend;⁴²² murdering ‘He is rioting in bouncing jumps and shattered forestry bursts, yet always inexorably tumbles away until he reaches even ground. Then he stops rolling even if he approached in a powerful manner. Accordingly Hektor also threatened to easily pass through murdering to the Danaian ships and tents as far as the shore of the sea.’

An interesting contrast is provided by examples from two different translations of Homer’s Odyssey. In the late 18th century, Voß employs the raising verb drohen to indicate an imminent event (cf. 502). In an earlier translation, more than 200 years earlier, Schaidenreisser uses another circumscription based on the adverb schier ‘almost’ (cf. 503). This could be an indicator that drohen with this particular function did not exist in the 16th century. (502)

Aber da eben jetzo der Ölbaumknittel im Feuer // Drohte zu but as just now the olive.pole in.the fire threatened to brennen, [...] zog ich ihn eilend zurück aus dem Feuer,⁴²³ burn-inf pull I it swiftly back out the fire ‘But as the pole of olive threatened to burn in the fire, I swiftly pulled it out of the fire again.’

(503)

Und da der oelbeümin pfal wol erhitzt war/ und schier and when the olive-adj pole well heated was and almost glueend worden⁴²⁴ glowing became

421 Friedrich Wilhelm Gotter Elektra, II, 2, Weimar (1772), ( printed in Gedichte, vol II, 45). 422 Illias, XIII, 140, translated by Johann Heinrich Voß, (1793). 423 Odyssee, IX, 140, translated by Johann Heinrich Voß, (1781).

2.2 Case studies |

247

‘And as the pole of olive was well heated and almost started to glow ’

These findings correspond to the results of the investigation carried out by Diewald and Smirnova (2010: 271). According to their view, the first unambiguous instances of drohen with a raising pattern can be found in the late 18th century. As Diewald and Smirnova (2010: 287) further point out, versprechen appears to have undergone the same development much later. They observe the first uses of versprechen with a raising pattern starting from the early 19th century. But this may be due to the fact that the raising pattern of versprechen is generally much harder to find. Turning to verheißen, the situation reflects the observations made by Diewald and Smirnova (2010: 287). Just as its semantic counterpart versprechen, the verb verheißen is found as a raising verb in the early 19th century. (504)

und möge in Hamburg, das einst für die vaterländische Bühne so and may in Hamburg that once for the Fatherland-adj stage so viel war und so viel wieder zu werden verheisst, nichts den much was and so much again to become promises nothing the guten Absichten eines Schröders in den Weg treten!⁴²⁵ good intention a-gen Schröder-gen in the way step-inf ‘And may there be nothing that prevents that the good intentions of Schröder become true in Hamburg which once was so important for the stage of the Fatherland and which promises to become it again.’

For all of the three verbs, the so-called uses with inanimate subject referents, as illustrated in examples (497)–(499), are an important landmark in their process of grammaticalisation. Before drohen, versprechen and verheißen could acquire their raising pattern, they had already developed non-agentive patterns without infinitive complements. As Diewald and Smirnova (2010: 271) illustrate, drohen is documented with non-agentive uses at least from Middle High German onwards. By contrast, versprechen could only be found in such configurations in the late 18th century, cf. Diewald and Smirnova (2010: 287). After having reviewed the origin and the characteristic of these three verbs, the question arises of how these patterns can be captured. A couple of analyses have been proposed. Some authors, e.g. Askedal (1997b: 14) and Wurmbrand (2001: 205), argue that the raising patterns of drohen and versprechen have to be considered as epistemic modal verbs. As Askedal (1997b: 14) argues, they behave like epistemic modal verbs in that (i) they always select obligatorily coherent infinitive complements, (ii) they fail to licence VP anaphora, (iii) they never occur

424 Odyssea IX, p. XXXIX, translated by Simon Schaidenreisser, (1537). 425 Morgenblatt für gebildete Stände N° 16, Freitag 18. Januar (1811), p. 64.

248 | 2 Modal verbs: A class struggle

in non-finite forms (bare infinitive, zu-infinitive, past participle), (iv) they cannot be used as a imperative, and (v) they do not impose selectional restrictions on their subject argument. However, on closer inspection, it turns out that the raising patterns drohen and versprechen (and verheißen) behave fairly differently from the traditional epistemic modal verbs. First of all, Askedal’s characterisation of these raising uses is not entirely correct. As Reis (2005: 140, 2007: 29) and Colomo (2011: 271) show, drohen and versprechen do not always occur in configurations that exhibit the word order that is crucial for obligatory coherence. Rather, they exhibit a pattern that could obviously be regarded as an instance of a pattern called “Third Construction”. Furthermore, Reis (2005: 133, 2007: 17) and Colomo (2011: 260–265) indicate that raising uses of drohen and versprechen occur with non-finite forms such as bare infinitives and past participles. Thus, the analogies between these raising verbs and the traditional epistemic modal verbs are by far less striking than Askedal (1997b: 14) suggests. Moreover, authors like Askedal (1997b: 14) and Wurmbrand (2001: 205) ignore a couple of essential differences between drohen and versprechen on the one hand, and the traditional modal verbs, on the other. As Reis (2005: 129, 2007: 13) and Colomo (2011: 241–245) demonstrate, they can readily occur in environments from which epistemic modal verbs are excluded or almost excluded, e.g. embedded under a past tense operator and in questions. Furthermore, Reis (2001: 312 Fn. 40, 2005: 130, 2007: 14) and Colomo (2011: 236) have shown that the raising uses of drohen and versprechen are restricted to the selection of predicates that refer to a future event. As will be demonstrated in Chapter 3, the essential characteristic of epistemic operators is the ability to embed predicates that refer to a present state or to an event in the past. Apart from that, it has been shown above (cf. 497– 499) that drohen and versprechen have uses that are semantically related to the raising patterns but do not involve infinitive complements. In contrast, epistemic modifiers always require some sort of clausal complements such as bare infinitives. Finally, drohen and versprechen always select zu-infinitives rather than bare infinitives. In contrast to brauchen, the infinitive particle zu cannot be dropped in any context. An alternative analysis has been proposed by Reis (2005: 140, 2007: 18), who considers the raising uses of drohen and versprechen as temporal-aspectual verbs. As she points out, they share a couple of characteristics with beginnen ‘begin’, anfangen ‘begin’, aufhören ‘stop’. More specifically, she shows that all of these verbs exhibit the Third Construction pattern to the same extent. While the traditional temporal-aspectual verbs are specified for the first or the last phase of a particular event, she demonstrates that drohen and versprechen describe the phase that immediately precedes that event.

2.2 Case studies |

249

This reasoning is plausible, as there are other temporal-aspectual verbs in German that exhibit a similar semantic specification, e.g. anschicken ‘to be about to’. However, it has been observed that the core class of temporal-aspectual verbs is rather reluctant to embed passivised infinitive complements, as has been documented by Haider (1993: 244), Reis (2005b: 135 Fn.8) and Colomo (2011: 290). By contrast, the raising uses of drohen and versprechen are occasionally found with infinitives that are passivised, as is illustrated in example (492). Does this indicate that these verbs cannot be considered as temporal-aspectual verbs? As it turns out, there is a group of temporal-aspectual verbs in German that can also be found with passivised complements. Interestingly, it is precisely the temporal-aspectual verb anschicken that is semantically very close to drohen and versprechen, as it is specified for the phase immediately preceding the event. (505)

Haiders FPÖ schickt sich an, erstmals in der Geschichte an Haider-gen FPÖ is.about refl on first.time in the history at einer österreichischen Regierung beteiligt zu a Austrian government participate-ppp to werden.⁴²⁶ pass.aux-inf ‘Haider’s FPÖ is about to participate in an Austrian government for the first time in history.’

(506)

Nun aber schicken sich Pavel Pardo und Ricardo Osorio an, in now but are.about refl Pavel Pardo and Ricardo Osorio on in den Annalen des Ländles verewigt zu werden.⁴²⁷ the annals the-gen Ländlegen immortalise-ppp to pass.aux-inf ‘But now, Pavel Pardo and Ricardo Osorio are about to be immortalised in the annals of the Ländle.’

The existence of examples like (505)–(506) is further support for the analysis developed in Reis (2005: 140, 2007: 18). Colomo (2011: 290) alternatively suggest that the raising verbs drohen and versprechen, together with scheinen and pflegen, which are also raising verbs, constitute the natural class of semi-modal verbs. However, as has been illustrated by Reis (2007: 17) and in Section 2.2.11, drohen and versprechen differ from scheinen in more respects than from the temporal-aspectual verb anschicken. At this point, it appears that Reis’ account is the one with the most explanatory power.

426 DeReKo: RHZ00/JAN.14751 Rhein-Zeitung, 27/01/2000. 427 DeReKo: HMP07/MAI.01447 Hamburger Morgenpost, 15/05/2007.

250 | 2 Modal verbs: A class struggle

2.2.13 Summary As has been indicated in Section 2.1.4, the six traditional modal verbs do not constitute a consistent and homogeneous class. In order to find out to what extent it is possible to provide an alternative classification, the previous sections have thoroughly investigated the following 14 elements, which are often considered as modal verbs in German (by various authors). (507) können, müssen, wollen, sollen, dürfen, dürfte, mögen, möchte, brauchen, werden, scheinen, drohen, versprechen, verheißen Based on the observation that there are not many verbs in German which could be considered as epistemic modal verbs, the availability of an epistemic interpretation became the guiding criterion in the preceding sections. Moreover, it turned out that epistemic modality can be characterised by means of two essential properties: (i) first of all, an epistemic operator indicates that the embedded proposition is not part of the deictic centre’s (speaker’s) knowledge (CoDeC). Secondly, (ii) epistemic operators can embed predications consisting of an identified subject referent and a predicate that refers to a state that cannot be changed or an event in the past. There are at least two ways of defining a class of modal verbs based on the concept of epistemic modality. Each of them has a different implication regarding the extension of the class. According to the stronger definition put forth by Öhlschläger (1989: 132), Engel (1996: 463), Diewald (1999: 1) and Reis (2001: 287), it is possible to define a class of modal verbs in German in terms of poly-functionality. This refers to the class of verbs which carry two different types of modality. (508) Strong definition of modal verbs A modal verb is characterised by the availability of a circumstantial modal interpretation and an epistemic modal interpretation. Corresponding to this definition, any verb which exhibits a circumstantial modal interpretation next to an epistemic one would be considered as a modal verb. Any verb which has only a circumstantial or an epistemic reading but lacks the other one would be banned from this class. Based on the observations made in the preceding Sections 2.2.1–2.2.12, the following distribution of modal readings emerges. 1. Circumstantial, epistemic and reportative: sollen, ?wollen 2. Circumstantial and epistemic: können, müssen, ?mögen 3. Circumstantial and marginally epistemic: brauchen 4. Only circumstantial: possibility modal dürfen, möchte

2.2 Case studies |

5. 6. 7. 8.

251

Only epistemic: weak necessity modal dürfte, werden, ?mögen Circumstantial and reportative: ?wollen Not epistemic: scheinen Not epistemic: drohen, versprechen, verheißen

In the case study on sollen in Section 2.2.6, it was illustrated that epistemic and reportative interpretations are by no means equivalent. Thus, they are carefully distinguished in the remainder of this section. Furthermore, it has been revealed that the subjunctive of the past forms dürfte and möchte have acquired a noncompositional interpretation and, thus, they have to be analysed as independent lexical elements. Moreover, the traditional preterite present mögen is constantly used less and less with infinitive complements and seems to disappear with certain ‘modal’ uses. At this point, it becomes possible to apply the concept of polyfunctionality to all of the potential candidates that come into consideration for a classification as a modal verb. In doing so, it turns out that the content of the resulting class of modal verbs is fairly different from what is generally expected. There are only two verbs that can be considered as polyfunctional without any restraint: können and müssen. Apart from that, brauchen exhibits a marginally developed epistemic interpretation. Therefore, it would be a rather marginal member of this group. In the case of mögen with an infinitive, it is not really clear to what extent it can still be interpreted in a circumstantial way. In contemporary standard written German, this type of interpretation seems to be absent. Aside from its circumstantial interpretation, sollen exhibits a very oblique epistemic interpretation, which is only available when the verb bears the morphology of the subjunctive of the past. In addition, it carries a reportative interpretation. Accordingly, it is not clear which status the reportative modality has for the classification. As has been shown in Section 2.2.6, reportative modality substantially differs from epistemic modality in semantic respect. Thus, it cannot be fully equivalent to epistemic modality. If reportative modality is seen to be irrelevant for the definition of modal verbs, then wollen cannot be regarded as a modal verb neither, unless one accepts that it involves an epistemic concessive pattern as well, as is suggested in Section 2.2.3. At any rate, there is a whole number of elements that would have to be excluded from the class of modal verbs: möchte lacks an epistemic interpretation and there is no consensus to what extent it involves a reportative one. Likewise, the possibility modal verb dürfen has no epistemic interpretation. Furthermore, there are verbs that do not have a circumstantial modal interpretation, such as werden and the weak necessity modal dürfte. As has been illustrated, the remaining verbs cannot be considered as epistemic as they do not fulfil both of the conditions for epistemic operators stated

252 | 2 Modal verbs: A class struggle

above: scheinen behaves unexpectedly with respect to the CoDeC, as the speaker occasionally knows that the embedded proposition is false. This is hardly ever the case with epistemic modal verbs that are inflected for the indicative. In contrast, drohen, versprechen and verheißen cannot embed any predication that refers to a present state or past event. As a consequence, these verbs will not be taken into consideration for the definition here. If the class of modal verbs is defined in terms of polyfunctionality, one needs to be aware that its extent will be much smaller than is generally expected: There are only two perfect members: können and müssen and a couple of half-hearted members: sollen, brauchen and mögen. Furthermore, this sort of definition in terms of polyfunctionality faces another challenge. Being based on the concept of circumstantial modality, this type of account has to provide a clear-cut definition for circumstantial modality as well. As can be seen, this is by no means a trivial matter. It is far from obvious to what extent it is possible to find a uniform definition for all of the different types that are generally subsumed under this concept: deontic modality, volitional modality, practical modality, the ability readings of können, the emotive readings of mögen. It should not be too surprising if these subtypes cannot be unified and if some of these subtypes needed to be excluded. The approach in terms of polyfunctionality implicitly suggests that all of the patterns subsumed under circumstantial modality share some essential characteristics. Yet, it has been illustrated in the preceding sections that each of the verbs investigated behaves in a fairly idiosyncratic way. As can be seen, the idiosyncrasies are dominant and it is not clear to what extent there is any feature that distinguishes them from all of the other verbs. In opposition to this, the concept of epistemic modality can be captured without too much ado. At this point, the question arises what the benefit of a definition of modal verbs in terms of polyfunctionality is from a descriptive perspective? Given the fact that it would not encompass all of the epistemic verbs, this definition would ignore the striking similarities between the epistemic verbs including werden and dürfte. First of all, they are semantically fairly homogeneous. Furthermore, they behave syntactically in a very uniform way: All of them select bare infinitive complements and all of them involve a raising pattern. Probably, it is more promising to abandon this attempt in favour of another approach. Alternatively, one could refrain from considering all of the non-epistemic patterns of the verbs under investigation and uniquely focus on their epistemic interpretations. In contrast to the strong definition, one could assume that a natural class of (poly-functional) modal verbs does not exist. Yet, it has been demonstrated that the epistemic interpretations of the verbs reviewed so far behave in a fairly uniform way. Accordingly, they can easily be grouped in a semantically and syntactically homogeneous class. This approach has a crucial advantage com-

2.2 Case studies |

253

pared to the definition based on polyfunctionality: It puts together all of the elements that indeed behave in an analogous way. Moreover, it is the precondition for any attempt to provide a principled explanation of the relation between the formal and functional peculiarities. In opposition, such an account does not make any statement about all of the non-epistemic uses. And it is fairly doubtful whether there is any fruitful generalisation to be made for these uses. (509) Weak definition There is no class of modal verbs but there is a class of epistemic modal verbs. If the two criteria for epistemic modal operators are acknowledged, this definition will result in the following extension: (510) kann, könnte, muss, müsste, sollte, dürfte, mag, braucht nicht, wird For the sake of precision, indicative forms are distinguished from the subjunctive of the past forms. As it turns out, they are also homogeneous from a syntactic perspective: All of them select bare infinitive complements and all of them involve a raising pattern. Likewise, Reis (2001: 308) already assumes that these syntactic properties appear to be a prerequisite for a verb in order to express an epistemic modality. This conspicuous connection could be revealing for any attempt to explain the nature of epistemic modality. Whereas this approach captures the epistemic modal verbs in a more systematic way, it leaves enough space for the diversity of the remaining non-epistemic uses. As is illustrated in the preceding sections, each of the potential modal verbs is ambiguous between several syntactic patterns. And there is no verb which exhibits the exactly the same set of patterns as any other potential modal verb: können is used as a transitive verb, as a control verb with event modification, as a raising verb with event modification and as a raising verb with propositional modification. In contrast, wollen occurs as a transitive verb, as a verb with finite dass-clause, as a control verb with event modification, as a negative polar raising verb with event modification, as a control verb with propositional modification and possibly also as a raising verb with propositional modification. Moreover, sollen can be used as a raising verb with event modification and as a raising verb with propositional modification with reportative or epistemic interpretation. A more detailed overview of the different complement types of each potential modal verb is illustrated in Table 2.5. As already pointed out by Lehmann (1995: 33) and Diewald (1999: 2, 34), these different patterns of a particular verb always represent different stages of its grammaticalisation: The transitive uses reflect the original use of the verb, the control verb uses a younger one and epistemic uses are the most recent ones. Accordingly,

drohen

versprechen

können mögen nicht mögen wollen nicht wollen möchte sollen sollte dürfen/ dürfte müssen nicht brauchen werden scheinen

vol –

– – – requ – –

– –

– – – – x x ( + NPdat ) –

–

vol

–

x ( + NPdat ) –

abil emot

predicative phrases – –

NPacc

x

–

– – – – – –

– –

–

– –

(mit-PP)

– – – – – x ( + NPdat ) x ( + NPdat ) x ( + NPdat )

vol –

dass -clause – emot emot vol

Tab. 2.5: Complement types of potential modal verbs

–

–

– – – – – –

– –

wenn -clause – (em), (vol) – –

–

– – – – – x ( + NPdat ) –

– –

(als) ob -clause – – – –

x ( + NPdat ) x ( + NPdat )

– – – requ – –

vol –

control inf. ev abil (vol) vol vol

–

–

– – – requ – –

vol –

control dir. ev abil (vol) vol vol

x

deoP – deoN, nec deoN, nec fut x ?( + NPdat ) x

? ext ? deoN

raising inf. ev deoP, pos ?

–

–

deoP – deoN, nec deoN, nec – – –

? ext ? deoN

raising dir. ev deoP, pos – ??

–

–

– – – – – –

x –

x

control inf. non ev – ?

–

–

– – – – – –

? ??? ? x

?

raising inf. non ev rep

–

–

? – x – x x x x ?

x

raising inf. non ev x x

254 | 2 Modal verbs: A class struggle

2.2 Case studies |

255

the diversity of these different readings reflects the history of the particular verb. As each verb has a different origin and an individual development, it is not surprising that there are hardly two verbs that have an identical set of patterns. In the case of the verbs considered above, the patterns become the more uniform the more grammaticalised they are. Each element takes up its individual path of grammaticalisation, and in the case of the verbs discussed above, these paths may converge in the end and, finally, they become epistemic modal verbs that select bare infinitive complements and that exhibit raising patterns. Related observations have been made by Lehmann (1995: 25). Therefore, it is much easier to provide a class definition that only considers the epistemic uses than a definition that encompasses other less grammaticalised uses as well. However, the most important finding that has been revealed by the case studies is the overwhelming diversity of different uses as has been documented throughout the preceding sections. This diversity of the non-epistemic uses poses a challenge for either of the approaches discussed above. Advocates of the strong definition would need to show that at least some of them can be considered as circumstantial modal uses. In turn, it becomes necessary to provide a clear cut definition of circumstantial modality in order to check which of these can be classified as circumstantial. It is fairly likely that there are some uses that cannot be classified as circumstantial uses such as the transitive uses or uses with finite dass- and wenn-clauses. As for these cases, advocates of the strong definition would need to come up with an explanation for all of those left-over uses. Likewise, the weak decision would need to account for those left-over uses as well. In contrast, it does not need to provide a definition of circumstantial modality that covers as many of the non-epistemic uses. Given the overwhelming multitude of different patterns and idiosyncrasies, any account needs to be flexible enough to capture this complex network of semantic relations. Specifically, it has to account for two facts. Most importantly, it has to explain how it is possible that all of the verbs under discussion are ambiguous between up to six semantically related patterns. As has been shown by Diewald (1999: 27) and Abraham (2003: 2), the relationships between the different patterns is a diachronic one. There are some patterns that have developed from others. Yet, even the most grammaticalised pattern of a modal verb retains the semantics of the original lexical meaning. This indicates that it should be possible to derive younger, more grammaticalised patterns always from the previous ones. Secondly, the approach has to explain the fact that each verb involves its individual selection of syntactic patterns. An analysis that captures all of the specific uses of one verb does not need to capture all of the uses of another verb. The most obvious approach to these challenges is one that allows for much variation and specification in the lexicon. A lexicalist theory such as HPSG has

256 | 2 Modal verbs: A class struggle

great advantages for this endeavour. Relations between the different patterns can be easily established by means of lexicon rules and type hierarchies. Summing up, it has been demonstrated that it is not a trivial matter to provide a definition of modal verbs that comprise all of the elements that are generally held as such. Even the promising attempt to define modal verbs in terms of polyfunctionality has turned out to be treacherous. First of all, it faces the burden of proving that there is a homogeneous class of circumstantial modality and that it captures all of the traditional modals. Moreover, it ignores the verbs that only exhibit an epistemic modal interpretation but no circumstantial one, such as werden and dürfte. Their epistemic uses would be left unexplained. At this point, the question arises what the benefit is of an account in terms of polyfunctionality. A less spectacular, yet more efficient solution can be attained based on the assumption that there is no homogeneous class of modal verbs. Instead, it can easily be shown that there is a homogeneous class of epistemic modal verbs. In contrast, this type of account does not have to make a statement on all of the non-epistemic uses. At this point, it is not clear whether there are any useful generalisations to be made for the non-epistemic patterns. Rather, it is sufficient to capture the lexical and semantic relations between each of the patterns. Thus, a homogeneous classifications is possible if only the epistemic modal verbs are considered. As will be shown in Section 6, it is even possible to derive the reportative uses from the epistemic ones by means of the CoDeC and some additional assumptions. Finally, it is more important to describe and capture all of the recorded forms, rather than postulating seductive over-simplified generalisations which do not match the empirical evidence. The establishment of a class of epistemic modal verbs is by far less spectacular than assuming the existence of a class of modal verbs. Yet, it is more thoroughly grounded in the data. Perhaps it is the term ‘modal verb’ that has caused all of this confusion. Possibly, it is just an invention that does not match reality. The next section will provide a brief overview of the history of the term ‘modal verb’ and its original motivation.

2.3 The origin of the term Modalverb In the course of the preceding sections, it has turned out that it is a great challenge to provide an intensional definition of the class of modal verbs. These difficulties could be evidence that the concept modal verb suffers from internal contradictions. Unfortunately, notions and terms are sometimes confused in the course of history; occasionally, they are reinterpreted in a less exact manner, sometimes they were not precise enough from the beginning. As Butt (2006: 153) illustrates, the well-known term ergativity has a spectacular history. Being misinterpreted,

2.3 The origin of the term Modalverb

|

257

the term ergative verb established in the early eighties has developed a meaning that is very far from the original intention. As it turns out, the term Modalverb has a remarkable development as well. The upcoming section will investigate the different definitions of verb classes that concern the six traditional modal verbs. The focus will be on the precise name of the class. Furthermore, the original motivation for the classification will be discussed. As it turns out, there were a number of different terms and conceptions in competition with each other. It is far from clear what precisely caused particular terms to prevail, whereas most of them were forgotten. Furthermore, it will be shown that it is not always the most consistent concept that remains in the scientific memory.

2.3.1 Early grammars: A morphological classification According to Jellinek (1914: 296), the term auxiliary was not used in Latin grammar. It has only been introduced by the French grammarian Pillot in his book Gallicae linguae institutio in 1550. Pillot (1550: 21, 24) argues that sometimes in French the use of a verbum auxiliarium becomes necessary in order to translate verbs with particular tense morphology from Latin. He discusses two verbs estre (’be’) and avoir (’have’). In much the same spirit, Ölinger (1574: 94) observes that these two auxiliary verbs sein (‘be’) and haben (‘have’) exist in German as well. Aside from that, he observes that werden has to be considered as a passive auxiliary in German. Ölinger (1574: 151) also mentions some of the verbs that would be known as modal auxiliaries later in history. As he notes, the five verbs woellen, sollen, doerffen, koennen and moegen behave unexpectedly in morphological respects. They will not be realised as a ge-participle whenever selected by a perfect tense auxiliary, but as an infinitive. Yet, he does not consider them as verbs with a particular grammatical function, or as auxiliaries. At this time, it were, in particular, morphological anomalies that had attracted the attention of scholars. In the same period, Claius (1578: 96) observed that there are nine verbs that lack suffixes in the 1st and 3rd person singular, which causes them to appear monosyllabic: können, mögen, woellen, sollen, wissen, taugen, thuerren, düerfen and müssen. Accordingly, he calls them verba monosyllaba. Occasionally, he employs woellen to circumscribe the Latin suffix for future tense. Yet, he does not associate these verbs with any particular grammatical function. Very much like Ölinger (1574: 94), the influential 17th century grammarian Schottel (1663:550) assumes there are three auxiliary verbs in German: seyn, werden and haben. Furthermore, Schottel (1663: 575, 579) notices that müssen, wollen, sollen, dürfen, können and mögen surface as a infinitives rather than past par-

258 | 2 Modal verbs: A class struggle

ticiples when they are embedded by a perfect tense auxiliary. Apart from that morphological peculiarity, he treats them together with the group of irregular verbs (ungleichfliessende Zeitwörter). What makes it particularly interesting to read Schottel’s work is that his own use of language differs from his descriptions. Accordingly, he uses two further verbs with IPP, which he does not mention in his enumeration: first, he uses the raising verb pflegen with IPP (cf. Schottel (1663: 243, 1019)) and second, he uses wissen with IPP (cf. Schottel (1663: 67, 144)). Bödiker (1698: 79), another popular and important grammarian, makes a revolutionary assumption. He concludes that two classes of auxiliaries have to be differentiated in German. Firstly, there are three merkliche Hülfwörter ‘memorable auxiliares’: sein, haben and werden. The second class is called gleichsam hülfwörter ‘quasi auxiliaries’: muessen, sollen, wollen, moegen, koennen, duerfen, wissen, cf. Bödiker (1698: 109). Most importantly, his definition is morphologically motivated:⁴²⁸ Drittens ist zu mercken/ daß insonderheit diese verba gleichsam auxiliara, muessen/ sollen/ wollen/ moegen/ koennen/ duerfen/ wissen/ ein doppelt participium haben auf et und en. Oder daß sie wenn ein ander verbum dazu koemmet/ als denn im perfecto keine praeposition ge haben/ sondern dem Infinitivo gleich außgesprochen werden. Als ich muß ich habe gemußt absolutè. Aber hergegen wenn ein ander verbum (in infinitivo) dazu koemmet/ so heisset es muessen. Ich habe muessen hoeren/ Ich habe muessen strafen. Also: ich kan/ ich habe gekont/ absolutè. Aber in dem ein ander infinitivo dazu koemmet so heist es koennen: ich habe koennen dencken. Er haette koennen sagen. Also auch: ich weiß/ ich habe gewußt/ absolutè. Koemt ein Infinitivus dazu/ so heisset es wissen. Er hat wissen zu sagen. Er hat wissen einzuwenden. Du hast es wissen zu verbergen.

Furthermore, Bödiker (1698: 101) acknowledges that pflegen ‘to be wont to’ exhibits the IPP as well, as soon as it is used with a foleo interpretation. However, it is not clear why he did not consider it as a ‘quasi auxiliary’, just as all of the other verbs that exhibit the IPP. Once again, the definition is clearly based on a morphological criterion. Yet, it remains mysterious why Bödiker (1698: 109) decided

428 Thirdly, it has to be noticed that these verbs being virtually auxiliara: muessen, sollen, wollen, moegen, koennen, duerfen, wissen. involve two different types of [past] participium, ending in et and en. Or whenever another verbum joins them, they will not have the praeposition ge in perfecto [tense] but will be pronounced like the Infinitivo, as ich muß ich habe gemußt absoluté. But whenever another verbum (in infinitivo) is attached it is realised as muessen. Ich habe muessen hoeren/ Ich habe muessen strafen. Alike: ich kan/ ich habe gekont/ absoluté. But as soon as another infinitivo is attached it is realised as koennen: ich habe koennen dencken. Er haette koennen sagen. Alike: ich weiß ich habe gewußt absoluté. If an Infinitivus is attached it is realised as wissen. Er hat wissen zu sagen. Er hat wissen einzuwenden. Du hast es wissen zu verbergen. [Own translation, the highlighted items are already contained in the original text.]

2.3 The origin of the term Modalverb

|

259

to treat these verbs as ‘quasi auxiliaries’ rather than as usual irregular verbs, just like the other irregular verbs are treated. Hence, an explicit motivation in terms of their function in grammar is missing. By contrast, Adelung (1782: 772) rejects Boediker’s approach. According to his view, auxiliaries are only those elements which are necessary to circumscribe verbal morphology in Latin, such as tense and passive suffixes. Correspondingly, he only considers seyn, haben and werden as genuine auxiliaries. Furthermore, he explicitly excludes ‘incomplete predicates’ from his concept of auxiliary, such as wollen, mögen, sollen, lassen, müssen, können:⁴²⁹ Nur solche sind Hülfswörter, durch deren Hülfe die zum Muster genommene Lateinische Conjugation umschrieben wird, und deren sind nur drey ’seyn’, ’haben’ und ’werden’, welche insgesammt Intransitiva sind und irregulär conjugiret werden.

‘Incomplete predicates’ are essentially characterised by the fact that they require a further verbal complement, an infinitive. Summing up, the early descriptions of the six traditional modal verbs can be characterised as follows. Most grammarians from the 16th until the late 18th century assume that there are a couple of auxiliaries in German. There is wide spread consensus about three verbs sein, werden and haben, which are generally regarded as auxiliary verbs. This classification is motivated by the fact that they become necessary in order to translate analytic tenses or passive suffixes from Latin into French or German. Implicitly, these authors assume that these verbs contribute grammatical information, such as temporal specifications, rather than lexical meaning. It is fiercely contested to what extent the traditional modal verbs should be regarded as auxiliaries as well. They have attracted attention due to their morphological anomaly. Yet, none of the authors who advocate a classification as auxiliaries provide a clear motivation for this. In this period, the six traditional modal verbs are not associated with a particular grammatical function. This does not change when Grimm (1822: 851, 1053) reveals the nature of this class, which attracted a lot of attention in morphological respect. As he illustrates, the singularity of the morphological class verba zweiter anomalie ‘verbs of the second anomaly’ is due to their origin: They reflect former preterite tense forms that have acquired present tense meaning. Furthermore, Grimm (1822: 853) has already noted that wollen does not belong to this class.

429 Only those verbs can be considered as auxiliaries that correspond to a morphem of inflection in Latin. There are three of those sey, haben and werden, all of them intransitive and exhibit irregular conjugation. [Own translation]

260 | 2 Modal verbs: A class struggle

Interestingly, Grimm (1822: 851) introduces the morphological class of verba zweiter anomalie in a small section that discusses auxiliaries. According to his view, they are additionally characterised in their high frequency:⁴³⁰ Auxiliaria, d .h. verba, welche sehr häufig gebraucht werden und statt ihrer lebendigen bedeutung abstracte begriffe annehmen, tragen gewöhnlich solche unregelmäßigkeiten an sich.

Even if Grimm (1822: 851) discusses the preterite presents and the former subjunctive of the past viljan in the context of auxiliaries, he does not attribute any particular grammatical function to them. Until this time, the traditional six modal verbs are only considered as a class in morphological respects. Usually, other preterite presents are also considered. Yet, these verbs are not associated with any specific grammatical functions, yet. Other overviews of the treatment of the six traditional modal verbs in early grammars were proposed by Jellinek (1914: 296), Öhlschläger (1989: 19–21) and Johnen (2006) with different results and focuses. Redder (1984: 303–329) and Öhlschläger (1989: 21) provide an extensive overview of the discussion in the 20th century.

2.3.2 Karl Ferdinand Becker (1836): From a morphological classification to a grammatical one The description of the auxiliaries in German suggested by Becker (1836: 174–186) is one of the crucial landmarks in the history of the term modal verb. Whereas former definitions were only motivated by their morphological anomaly, Becker associates these verb with a particular grammatical function and justifies their status as auxiliary verbs. Moreover, he is the first author who tries to provide a systematic semantic description of this class. In doing so, he discovered a whole range of phenomena. A couple of years ahead of Becker, Heyse (1822: 402) makes the first attempt to provide a functional motivation for the definition of a modal auxiliary like class. As he assumes, there are various classes of auxiliaries in German. Among them, there are seven Huelfsverba welche den Modus umschreiben ‘auxiliaries which circumscribe mood’: dürfen, können, lassen, mögen, müssen, sollen, wollen. A further criterion for his concept of auxiliary-hood seems to be their requirement for a fur-

430 Auxiliaries, that is verbs which are very frequently used and which assume abstract concepts rather than their vital meaning, usually exhibit such anomalies. [translation JM]

2.3 The origin of the term Modalverb

| 261

ther infinitive complement, as illustrated in Heyse (1822: 403). Yet, he does not become very explicit in this matter. By contrast, Becker (1836: 176 §91) explicitly argues that every language has its proper Huelfsverben des Modus ‘auxiliaries of mood’. Their function is to express the possibility or necessity of the embedded predication. In other words, any modal verb is either a verb that expresses a possibility or a necessity. Furthermore, Becker (1836: 178) concludes that there are three different types of possibilities and necessities that an auxiliary of mood can express: physical, moral and logical. In his main classification, he considers können, dürfen and mögen as possibility verbs and müssen, sollen and wollen as necessity verbs. Apart from that Becker (1836: 182, 1841: 222) explicitly regards the AcI verb lassen as an auxiliary of mood, as it can express both a necessity (causative) reading and possibility (permissive) reading. Apart from that, Becker (1836: 177 §91) explicitly notices at some earlier point that brauchen is an auxiliary of mood. For some reason, he did not consider it in the general classification in §92 and §93. Interestingly, it appears in the second revised edition in this classification, in which it is treated as verb of a moral necessity, cf. Becker (1842: 220 §91, 224 §93). The morphological aspect appears to be less important for Becker’s definition. There are only scarce references to the morphological peculiarity of the verbs he considered. At one point, Becker (1836: 176 §91) remarks that the auxiliaries of mood display an anomalous conjugation pattern, which is due to their development: They are past forms that have acquired present tense meaning. In doing so, he is the first grammarian that does not focus on the morphology of these verbs. The fact that Becker (1836) does not mention their morphological anomaly should arouse suspicion. And indeed, in ignoring the morphological nature of the six traditional modal verbs, he avoids a couple of contradictions that would become apparent otherwise. First of all, he considers two verbs as auxiliaries of mood that neither have a preterite present origin, nor adopt a corresponding morphological paradigm: lassen and brauchen, as is illustrated in Becker (1836: 177 §91, 1836: 182, 1841: 222, 1842: 220 §91, 224 §93). Apart from that, his definition also faces challenges in semantic respect. Building on the assumption that modal verbs express possibilities and necessities, Becker (1836: 178) cannot properly capture epistemic duerfte, which he considers as a verb that expresses a probability. In his definition, Becker adopts a view that is fairly exceptional for a grammarian of his time. This could be due to the fact that he was not educated as a grammarian, but rather as a physician, who only developed his linguistic interest when he was older than 30. Even if his definition may suffer from inconsistencies, he nevertheless contributed a couple of ground-breaking discoveries, which enabled him to provide a description of these verbs that was much more efficient than those put forth by his contemporaries.

262 | 2 Modal verbs: A class struggle

Likewise, Heinrich Bauer (1827: 93) assumes in the first volume of his grammar Vollständige Grammatik der neuhochdeutschen Sprache that there are various types of Hülfszeitwörter ‘auxiliaries’ in German. Aside from haben, sein and werden, the incomplete predicates such as können, sollen, müssen can also be considered as such. He becomes more explicit in the third volume of his grammar. Following the work of Becker, Bauer (1832: 489–492) concludes that there are seven verbs that can be considered as Huelfsverben des Modus. Another important definition has been developed by Schoetensack (1856: 269, 293). Apart from the consensual auxiliaries haben, sein and werden, there are various verbs which he considers as auxiliaries as well. According to Schoetensack (1856: 291), bleiben ‘stay’, stehen ‘stand’, kommen ‘come’ and gehen ‘go’, have to be treated as auxiliaries as well. Furthermore, he suggest to consider the AcI verbs and the temporal-aspectual verbs as auxiliaries too. In most of the cases, he does not provide a grammatical or functional motivation for his classification. Accordingly, his concept of auxiliary remains fuzzy. As can be seen, the selection of infinitive complements plays an important role for his definitions. In a section on irregular verbs, Schoetensack (1856: 267, 269) regards six verbs as Huelfsverben ‘auxiliaries’ können, wissen, mögen, sollen, müssen, dürfen. These verbs are identical to the remnants of the class of preterite presents, which are still in use in New High German, including wissen. As he emphasises, wollen is not part of this class, as it has not undergone the same development. Seemingly, the driving force behind this definition is one that is based on diachronic morphology. In a subsequent section, Schoetensack (1856: 293) introduces the class of modale Hülfsverben ‘modal auxiliaries’ comprising seven verbs: mögen, wollen, können, sollen, müssen, dürfen and lassen. After the nomenclature defined by Becker (1836: 176), that is a further step towards the contemporary term ‘modal verb’. In opposition to the previous classification, wissen is excluded from the class and wollen and lassen are integrated. Yet, Schoetensack (1856: 297) acknowledges that lassen belongs to a different inflectional pattern than most of the other ‘modal auxiliaries’. Accordingly, Schoetensack’s term ‘modal auxiliary’ appears to be motivated in a semantic or grammatical way. Yet, he does not become very explicit about this matter. In another context, Schoetensack (1856: 558) remarks that modal auxiliaries in German are sometimes used to translate and substitute mood. But they do not systematically replace mood, since they also inflect for the subjunctive. This could be a hint that his term ‘modal auxiliary’ is based on a functional motivation. Apart from that, Schoetensack (1856: 295) considers the verb brauchen as an ‘auxiliary’ that is used to negate sollen. At this point, it remains unclear whether he considers it as a ‘modal auxiliary’ or some other auxiliary. Finally, Schoetensack (1856: 298): discusses the IPP. In this context, he uses the more general term Zeitwort ‘verb’ rather than Huelfsverb ‘auxiliary’. This group also contains elements which he has char-

2.3 The origin of the term Modalverb

| 263

acterised as auxiliaries in a previous section: hören, heißen, sehen, helfen, lassen, sollen, wollen, mögen, dürfen, müssen, wissen, können, fühlen, lehren and lernen. As has been shown, Schoetensack (1856) uses the term ‘auxiliary’ in an inflationary manner. In most cases, it is not evident what particular characteristic he associates with it. At some point, he seems to motivate the class of modal auxiliaries in a functional way. This class is almost identical with the six remaining preterite presents including wissen, which form a homogeneous group in morphological respect, and which he only considers as ‘auxiliaries’. At this point, it remains mysterious how the class of preterite presents and the class of ‘modal auxiliaries’ relate to each other, according to the view taken by Schoetensack (1856). Moreover, it is evident that his conception suffers from the same conflict as Becker’s definition: on the one hand, he tries to motivate the class in terms of morphological characteristics and on the other hand, he suggest a semantic definition of the class based on the concepts necessity and possibility. Yet, the two motivations do not converge. The morphological part of this definition yields another extension than the semantic part. Much in the spirit of Becker (1836) and Schoetensack (1856), Vernaleken (1861: 94), a grammarian with descriptive ambitions, postulates a class of modale Hilfsverben ‘modal auxiliaries’ which encompasses six elements: sollen, müssen, mögen, können, dürfen, wollen. His definition deserves closer attention, as he uses a term that is almost identical to the contemporary use and moreover he suggests an extension that precisely corresponds to the traditional six modal verbs. As he argues, the term Modalität ‘modality’ refers to the morphological anomaly of these verbs and their ability to express a Nothwendigkeit, Möglichkeit und Zulässigkeit des ausgesagten ‘necessity, possibility and permissibility of a proposition’. As this indicates, Vernaleken’s approach is a strong simplification. In opposition to his two predecessors, he ignores the semantically related verbs brauchen and lassen in his description. Apart from that, it remains to be shown that these six verbs fulfil his semantic requirement. Before Vernaleken, Friedrich Bauer (1850: 102 §166), whose grammar is the predecessor of the Duden grammar, already suggested a class of Hilfszeitwörter ‘auxiliaries’, which comprises the six traditional modal verbs wollen, können, mögen, sollen, müssen and dürfen. Yet, Bauer (1850: 30 §66) is by far less explicit in his motivation and seems to be primarily interested in their irregular morphology and refers to them as abweichende Verben ‘deviate verbs’. Yet, he appears to have a functional criterion in mind as well as he explicitly compares them to werden, which is employed to circumscribe future tense.⁴³¹ Der Inf. mit werden steht zur Umschreibung des Fut. In ähnlicher Weise steht er bei den Hilfszeitwörtern wollen, können, mögen, sollen, müßen, dürfen.

264 | 2 Modal verbs: A class struggle

Beginning with Becker (1836), definitions of modal auxiliaries appeared that were not solely grounded on morphological aspects. Some of the authors of that period associated these morphologically anomalous verbs with a very homogeneous meaning. According to their assumption, each of the verbs either expresses a necessity or a possibility. As it turns out, the set of verbs with preterite present morphology is not congruent with the set of the verbs that express a necessity or a possibility. All of these accounts that are based on this double motivation suffer from the same inconsistencies. Most of the authors ignore the apparent contradictions. The increasing importance of the semantic motivation is reflected by the nomenclature that has been introduced by Becker (1836). Subsequently to his ground-breaking work, the class is called ‘auxiliary of mood’ and later ‘modal auxiliary’. These terms almost correspond to the standard term used in contemporary grammars of German: Modalverb.

2.3.3 Blatz (1896) In the late 19th century, it is generally accepted that there are more auxiliaries than haben, sein and werden. Most of the grammarians assume that there is at least one other class, which is called modal auxiliaries or auxiliaries of mood. In this period, a further interesting development took place in the late 19th century. When Blatz (1896) published the third edition of his grammar, he discusses a class of verbs that he refers to as Modalverba ‘modal verbs’. According to the view held in Blatz (1896: 538 §154), this class encompasses the five preterite presents können, mögen, dürfen, müssen, sollen and the two verbs wollen and lassen. In a brief remark, Blatz (1896: 542) also considers brauchen as a Modalverb. According to his selection of verbs, he favours a definition that is semantically motivated. Moreover, he states that they have a similar function as grammatical mood, yet, they are more specific in their interpretation. This deserves closer attention, as he does not use this expression in the preceding edition of his grammar. In the second edition, Blatz (1880: 649 §396) chooses the term Hilfsverben des Modus ‘auxiliaries of mood’ and in an earlier section Formzeitwörter der Aussageweise, which corresponds to a ‘germanised’ version of the original Latin terms, cf. Blatz (1880: 267). As can be seen, something made Blatz change his mind between the year 1880 and 1896, which lead him to using the term Modalverb. Yet, it remains fairly mys431 The infinitive with werden is used to circumscribe future tense. In a similar manner, it is used with the auxiliaries wollen, können, mögen, sollen, müßen, dürfen [own translation]

2.3 The origin of the term Modalverb

| 265

terious what incidence or influence this could have been. One of the major suspects is of course Henry Sweet, who published his influential New English Grammar. Logical and Historical in 1891. As the subtitle indicates, the investigation is grounded in an explicit logical interest. This leads to the conclusion that he could be the origin of a new terminology that follows the tradition of by (modal) logics. Yet, Sweet (1891: 421) turns out to be rather conservative in his nomenclature as far as auxiliaries are concerned. First of all, he is not very consistent in his choice, he alternates between the terms ‘anomalous verbs’ and ‘preterite presents’ (can, dare, may, must, ought, shall, will) acknowledging that need also fits into that paradigm, having lost its 3rd P -s and taking the negation particle not. At another point, he employs the term ‘chief auxiliaries’ referring to the elements be, have, do, will, shall and may. The latter, he characterises as auxiliary of permissive mood as in ‘May you be happy!’. In a preceding section, Sweet (1891: 108) discusses various instances of periphrastic mood and the corresponding auxiliaries such as would, the auxiliary of conditional mood should, the auxiliary of compulsive mood to be+INF and the auxiliary of permissive mood may. As this indicates, Sweet (1891) does not use a terminology that is influenced by modal logic and he is certainly not the one who has inspired Blatz in his nomenclature.

2.3.4 George O. Curme Another grammarian who takes a related approach is the American George O. Curme. Decades after being a visiting scholar in Berlin in 1896, he published his influential A Grammar of the German Language. Curme (1922: 317) introduces a class of ‘auxiliaries of mood’ that encompass the six traditional modal verbs: dürfen, können, mögen, müssen, sollen and wollen. Even if it is homogeneous from a morphological perspective, Curme (1922: 318) is aware that wollen is not a real preterite present and that there is a further preterite present that is not part of this class: wissen. As he argues, the latter cannot be considered as a ‘modal auxiliary’, as it does not exhibit the IPP-effect, and as it fails to embed bare infinitive complements. In its extension, Curme’s class is analogous to the one defined by Vernaleken (1861: 94). Yet, Curme (1922) is inconsistent in his use of the designation: sometimes, he refers to this verb class as ‘past present verbs’, sometimes as ‘auxiliaries of mood’ and sometimes as ‘modal auxiliaries’. Apart from that, Curme (1922: 318) delivers a systematic description of each individual verb, which includes discussions of the epistemic uses of dürfte, können, mögen, müssen and the reportative use of wollen and sollen. This characterisation, by and large, reflects the one that was provided by Becker (1836), except that lassen and brauchen are not considered.

266 | 2 Modal verbs: A class struggle

Curme (1922: 318) does not become very explicit as far as the motivation of his definition is concerned. Considering the extension of his class, it appears that morphological reasons are dominant. This is further reflected by the fact that the modal auxiliaries are discussed in the context of irregular verbs. Yet, Curme seems to assume that the class can be motivated on the basis of independent functional or grammatical reasons. This becomes more evident in his English Grammar when Curme (1931: 393 ) discusses the function of modal auxiliaries in English: As subjunctive forms lost their endings modal auxiliaries were pressed into service to express the same ideas [...] they ceased to be verbs and are now in reality mere grammatical forms to color the statement.

In this statement, the function of modal auxiliaries in English is identified as a means to express mood. But as English differs from German in that respect, one should avoid premature conclusions about what the functional motivation of Curme’s definition of German modal auxiliaries could be.

2.3.5 Braune, Paul and Behaghel: The decline of the semantic motivation At the same time, a new tradition arises among the positivist Neogrammarians, focusing on the morphological aspects of the verbs under discussion here, rather than functional or grammatical ones. Their particular interest might be due to their explicit historical interest. In his Old High German grammar, Braune (1886: 252, 259) limits himself to the morphological development of the praeteritopraesentia and wollen, which developed from a former subjunctive of the past form. He does not make any comment about a semantic or functional motivation. This might be partially caused by the fact that most of the traditional six modal verbs behaved in a different way in Old High German. Moreover, historical semantics is always a very speculative endeavour, too speculative for the positivist Neogrammarians. Some of the subsequently published grammars of New High German follow this tradition, e.g. Wilmanns (1906: 92–99 §51–§55), Sütterlin (1910: 232, 240) and Paul (1917: 262 §190–§195 ). The three of them adopt the term ‘preterite present’ and they confine themselves to the morphological anomalies of this class. Apart from that, they discuss the development of the former subjunctive of the past wollen. In a similar way, Sanders (1908: 222) focuses on the formal peculiarities of these verbs. In contrast to the other grammarians discussed here, he does not consider the traditional modal verbs as a separate class. Rather, he assumes that they constitute a class together with all of the verbs that exhibit the IPP in German.

2.3 The origin of the term Modalverb

| 267

According to his view, all of these verbs should be regarded as auxiliaries: dürfen, heißen, helfen, hören, können, lassen, lehren, lernen, machen, mögen, müssen, sehen, sollen, wollen and occasionally brauchen, pflegen, suchen and more rare empfinden, erblicken, finden, fühlen, schauen, wissen and zeigen. Similarly to the other approaches sketched here, he does not provide any functional or grammatical motivation. Finally, Behaghel (1924: 309) does not pay so much attention to this group of verbs. In a section on auxiliaries, he refers to them as Hilfsverben, die ein Dürfen, Können, Sollen, Wollen bezeichnen ‘auxiliaries that refer to a permission, possibility, obligation or a wish’. Interestingly, his semantic characterisation is neither exhaustive nor systematic. Furthermore, he does not provide any motivation for his classification. In a subsequent section, Behaghel (1924: 368), once again, employs the term Hilfsverben ‘auxiliaries’ dürfen, können, mögen, müssen, sollen, wollen, when he notes that they all exhibit the IPP-effect. Furthermore, he acknowledges that brauchen manifests the IPP-effect as well. However, Behaghel does not become explicit about the nature of this verb. Likewise, the term Hilfszeitwort ‘auxiliary’ is used for these verbs in the Trübners deutsches Wörterbuch edited by Götzke and Mitka (1939). Most of these definitions that evolved in the early 20th century refrain from providing a grammatical or functional motivation. Either they are solely motivated on morphological grounds, or the grammatical motivation remains implicit. There are analogies to the research on modal auxiliaries in English at that time. Jespersen (1917: 92–96) enumerates the relevant elements yet without labelling them with a specific term. At some later point Jespersen (1917: 94) refers to the pattern mustn’t as a ‘prohibitive auxiliary’. In the late thirties, the foundations were laid for a grammar which became the most popular one after World War II, when Basler (1935) edited the first Duden grammar. It is basically grounded on Friedrich Bauer’s Grundzüge der neuhochdeutschen Grammatik für höhere Bildungsanstalten, which was published in 1850. After 17 successful editions in thirty years, it was revised by Konrad Duden and republished in 1881, as the 18th edition of Grundzüge der neuhochdeutschen Grammatik für höhere Bildungsanstalten und zur Selbstbelehrung für Gebildete. As already demonstrated above, Bauer (1850: 30 §66, 102 §166) assumes that there is a class of auxiliaries in German that consists of the six traditional modal verbs. In this respect, Bauer’s description did not undergo any substantial change, at least until the ‘12th edition for catholic schools’, cf. Bauer (1870: 64 §66, 163 §167). However, in a footnote he refers to single verbs as Hilfszeitwörter des Modus ‘auxiliaries of mood’, cf. Bauer (1870: 158 §165 Fn.). Likewise, the subsequent version edited by Bauer and Duden (1887: 69 §66) adopts, in essence, the descriptions given in earlier editions. There are only minor revisions concerning the termino-

268 | 2 Modal verbs: A class struggle

logy. In the section on morphology, these verbs are called unregelmäßige Verba ‘irregular verbs’, rather than abweichende Verben, as in the earlier editions. Furthermore, Bauer and Duden (1887: 69 Fn.1) notice in a small footnote that these verbs plus the verb wissen constitute a class that is called Praeterito Praesentia and that essentially exhibits the IPP. In his terminology, Basler (1935: 89) avoids Latin expressions and, accordingly, chooses a denomination that is purely German Hilfszeitwörter der Redeweise. In contrast to the original predecessor Bauer (1850: 102 §166), he suggests a class extension which contains the six traditional modal verbs including lassen: können, mögen, wollen, dürfen, sollen, müssen and lassen. A considerable criterion for the classification proposed by Basler (1935: 94) seems to be the availability of the IPP-effect. Yet, he acknowledges that there are two more verbs which are found with this pattern: heißen and sehen. Moreover, he remarks that hören, helfen and lernen are also found with IPP-morphology; yet, in this cases, this is the result of a misunderstood assimilation (‘In falsch verstandener Angleichung’). Basler (1935) obviously adopted this view from Bauer and Duden (1887: 69 Fn. 1), who provide an almost identical formulation. However, their evaluation deserves attention, as it is in conflict with the evidence provided by Kurrelmeyer (1910), who demonstrated that precisely hören was one of the first verbs which exhibited the IPP in Late Middle High German. Moreover, Basler (1935: 111) argues that they are preterite presents except for lassen. As has been demonstrated, Neogrammarians return to a consistent definition. This is due to their particular interest in phenomena which can be clearly observed. Accordingly, their classifications are based on purely morphological grounds. The functions of the relevant verbs are largely ignored. This is also reflected in the term that is employed in these grammars. The expression praeterito praesentia refers to the morphological development of these verbs, rather than their function or role in grammar. This perspective was adopted by several of the major grammarians in the early 20th century. Other authors remain very superficial and intuitive, as far as the grammatical motivation of their definition is concerned. Terms that contain a reference to mood or modality become rare again.

2.3.6 Gunnar Bech (1949) It is a term-paper that Gunnar Bech wrote when he was a student that later became on the maybe most influential contribution to the contemporary reasoning on modal verbs in German after it has been published in 1949.⁴³² The study entitled Das semantische System der deutschen Modalverben ‘The semantic system of the German modal verbs’ is frequently cited for a couple of achievements. Firstly,

2.3 The origin of the term Modalverb

| 269

Bech (1949) has reintroduced the term Modalverb, and second, he pleads for a class of modal verbs that comprises the six verbs wollen, sollen, dürfen, mögen, müssen and können. He is the first one to combine this particular term with the six modal verbs. As the title already indicates, Bech’s main interest is the semantics of these six verbs. Although they exhibit an analogous morphology, this fact does not receive much attention in his study. His major objective is to argue that these six verbs together constitute a ‘semantic system’. In these respects, he can be regarded as the father of the traditional six modal verbs in German. As with many studies on grammar in earlier times, Bech (1949) has not documented his references to other studies. Correspondingly, it remains fairly speculative which parts of his investigations are of his own achievement and which adoptions from other surveys. Yet, it is not very likely that he, being still a student, reinvented the wheel in every detail. Moreover, he was supervised in Copenhagen by a couple of influential scholars, such as the structuralist Louis Hjelmslev, and it is not clear to what extent this term-paper reflects the perspectives of his teachers. Finally, there are four striking parallels to various preceding studies most notably to the ones that have been carried out by Becker (1836), by Schoetensack (1856), by Vernaleken (1861: 94) and by Blatz (1900). As far as the name of the term is concerned, there are not too many influences that come into consideration. Prior to Bech’s work, only a few more investigations can be found that have used the precise term Modalverb: The grammar written by Blatz (1896) and Rossmann (1908). Similar terms, such as ‘modal auxiliaries’ or ‘auxiliary of mood’, are employed by Curme: 318, (1922: 318, 1931) and Kirchner (1940). It is fairly likely that Bech was familiar with Curme’s grammars, as he explicitly quotes them in a subsequent study, cf. Bech (1963: 292). Yet, it remains mysterious what precisely caused him to adopt this term that had previously not been used very frequently. Moreover, there are similarities with respect to the extension of the suggested class. As has been illustrated above, classes that encompasses the six members are not very frequent. Previous authors who have suggested an extension of this type are Bauer (1850: 102 §166), Vernaleken (1861: 94) and Curme (1922: 371). Apart from that, the extensive case studies presented in Bech (1949) are strikingly reminiscent of the descriptions provided by Becker (1836) and Schoetensack (1856). This concerns for instance the analysis of sollen as a verb of external volition.

432 A short biography of Gunnar Bech is provided as a preface of the reprint of Bech’s (1955/57) Studien über das deutsche Verbum infinitum published by Niemeyer in Tübingen in 1983.

270 | 2 Modal verbs: A class struggle

Finally, the way Bech (1949) motivates his definition is by and large analogous to the one put forth by Becker (1836: §90–§93). The latter argues that all of the eight verbs he has considered express either a necessity or a possibility, including the volitional verb wollen, which he analyses as a verb that expresses an inner necessity. In contrast to Becker, Bech (1949) only considers six verbs as relevant for his definition, leaving aside lassen and brauchen. Moreover, Bech (1949: 38) arranges them in a structuralist feature matrix consisting of a couple of feature oppositions such as ‘necessity/possibility’ and ‘subject referent is identical/not identical to the modal source’. This matrix should justify the existence of this modal verb class consisting of its six members. In essence, Bech’s analysis is Becker’s description in a structuralist guise. Furthermore, Bech (1949) managed to establish a concept that seemed both plausible and simpler than most of those that were preceding – even if it might suffer from essential contradictions and inconsistencies. Yet, it remains mysterious what exactly made him adopt the rare term Modalverb and what caused him to abandon the verbs brauchen and lassen, which have originally been considered as auxiliaries of mood. In his Kurze deutsche Syntax, Dal (1952: 107) does not provide an extensive discussion of the modal verbs. He contents himself with mentioning that there are six modale Hilfsverben that precisely correspond to the selection made by Bech (1949). In a subsequent section, Dal (1952: 145) briefly remarks that these verbs can be used to circumscribe subjunctive of the past. Much in the spirit of Bech (1949), Welke (1965) advocates a class of ‘Modalverben’ that comprises the same six verbs as Bech suggested. In contrast, Welke (1965: 11) provides a classification in terms of syntactic characteristics. According to him, the traditional six modal verbs are those verbs that (i) obligatorily select bare infinitive complements, (ii) require subjects of the infinitive that are co-referent with the matrix subject and (iii) exhibit no semantic restriction with respect to their infinitive complement. As he argues, this characterisation yields exactly the traditional six modal verbs. In a brief remark, he acknowledges that this definition would also include tun and werden. Finally, Welke (1965: 12) concedes that any syntactic classification comprising the traditional six modal cannot be achieved without making arbitrary decisions. However, Welke’s investigation happened to support and strengthen Bech’s analysis in the end.

2.3.7 The Duden grammar For some mysterious reason, German grammars that have been published after World War II seem to suffer from amnesia. The most popular ones edited by Grebe (1959: 83) and by Griesbach and Schulz (1960) adopt the term modal verb and as-

2.3 The origin of the term Modalverb

|

271

sume that this class only encompasses six items. In opposition to its immediate predecessor Basler (1935: 89), Grebe (1959: 83) does not consider lassen as a modal verb anymore. As none of these grammars makes an explicit reference in the respective sections on modal verbs, one can only speculate about possible influences. Yet, the precise choice of the terms and the class extension could be very revealing. In previous studies, the term and the class extension have not been used very frequently. As Bech (1949) and Bech (1951) was the first who combined both of them, it seems to be fairly likely that he is the major reference for the two grammars. For later grammars, the first comprehensive monograph on modal verbs published by Welke (1965) becomes another influential point of reference. In the first edition of the Duden grammar edited by Grebe (1959: 83), the denomination of the verb class is not yet consistent. In an earlier section on auxiliaries, they are referred to as modifizierende Verben ‘modifying verbs’. Later, in the major section dedicated to these verbs they are introduced as Modalverben, which comprise the traditional six verbs, cf. Grebe (1959: 485). Finally, Grebe (1959: 486) remarks that brauchen can be used in a similar fashion. The entire treatment of the modal verbs does not exceed a couple of small paragraphs in this first edition. In the second edition of the Duden grammar edited by Grebe et al. (1966: 528), the sections on modal verbs have not undergone any substantial changes except for the observation that brauchen exhibits the IPP-effect when used with a infinitive complement, cf. Grebe et al. (1966: 131) referring to Kolb (1964). In opposition to that, the third edition of the Duden grammar edited by Grebe et al. (1973: 66) contributes a comprehensive section on modal verbs, in which the term Modalverben is systematically used. Once again, it is assumed that the class comprises the traditional six members. Following the spirit of Welke (1965), a couple of characteristics are given in which the modal verbs are claimed to differ from all of the remaining verbs: (i) they select bare infinitive complements, (ii) they are preterite presents, (iii) they exhibit IPP, (iv) they can select the infinitive of the perfect and (v) they convey an additional meaning. Yet, the authors concede that a clear-cut delimitation is not possible. Finally, Grebe et al. (1973: 72) makes mention of the epistemic modal interpretation for the first time. In a similar manner, the fourth edition edited by Drosdowski et al. (1984: 94– 105) has undergone a considerable revision. The section dedicated to the modal verbs has been extended from a couple of paragraphs in the first two editions to 12 pages in the fourth edition. Hermann Gelhaus, the author of that section, reveals all of his sources encompassing the work of Bech (1949), Kolb (1964), Welke (1965), Fourquet (1970) and other research literature from that period. Drosdowski et al. (1984: 94) characterise modal verbs as verbs that modify the content of another verb. Moreover, it is argued that (i) they select bare infinitive complements, (ii) they cannot occur as imperatives and (iii) they withstand passivisation. Apart

272 | 2 Modal verbs: A class struggle

from the traditional six modal verbs, brauchen receives attention as well. It is considered as a close relative of the modal verbs. Finally, Drosdowski et al. (1984: 112) notice that lassen + sich is a pattern that concurs with the modal verbs. In the fifth edition edited by Drosdowski et al. (1995: 92–104), there is once again a proper section on a class of modal verbs, which encompasses the traditional six members. The content has not undergone any substantial changes with respect to the preceding edition. Once again, the seventh edition edited by Eisenberg et al. (2005: 562) was subject to comprehensive revisions. The main article has been written by Cathrine Fabricius-Hansen, who makes explicit reference to Bech (1949), Kratzer (1981), Brünner and Redder (1983) and Zifonun (1997). In contrast to the previous edition, the content has been reduced and is presented in a more condensed form. Whereas the designation of the class has remained the same, there are some minor changes in the extension. At the outset, the traditional six modal verbs are only mentioned. However, in the subsequent lines brauchen is explicitly considered as a modal verb. Moreover, Eisenberg et al. (2005: 566) notice that möchte can be regarded as an independent item as it is occasionally used with a proper infinitive. Eisenberg et al. (2005: 563) provide a description that has a semantic focus. As the authors conclude, modal verbs are characterised by three essential modal dimension: modal force, modal base and the modal source. By contrast, the grammar edited by Griesbach and Schulz (1960: 41, 65) is grounded on different assumptions. According to their view, the class of modal verbs is characterised by the selection of bare infinitives and availability of two modal interpretations: modal and epistemic. Thus, Griesbach and Schulz (1960: 41, 65) provide a definition that is functionally motivated. It is the first grammar that stresses the singularity of the epistemic modal interpretations. Yet, they conclude that the class only encompasses the six traditional members. As this section has revealed, the two popular grammars edited by Duden and Griesbach and Schulz (1960) have employed the term Modalverb since the early sixties. As they have used it throughout the present, it is fairly likely that they have played an important role in the proliferation of this expression that was almost not in use before Bech’s (1949) monograph had been published. In a similar fashion, Redder (1984: 306) has shown that the grammars which have been published after the second world war differed with respect to the term they employed: Modalverb or modales Hilfsverb. The former term could only prevail in the seventies.

2.3 The origin of the term Modalverb

|

273

2.3.8 Summary As the preceding sections have demonstrated, the German concept of Modalverb suffers from an essential inconsistency. According to a view held by most contemporary authors, the class should be motivated by two different types of criteria: by morphological criteria and by functional or grammatical criteria. However, the set of verbs that exhibit the morphology typical of modal verbs is not co-extensive with the set that displays modal semantics. It was widely observed in the 16th century that some of the verbs which are known today as modal verbs behave in an awkward way. Accordingly, it has frequently been suggested that they form a class that has to be treated separately from the other verbs. Yet, the precise term and the extension of this class differs from author to author across the centuries, as is illustrated in Table 2.6. Being the more salient feature, it was initially the morphological anomalies that attracted the attention of grammarians. In the descriptions that were published prior to those provided by Becker (1836: 174–186), the definitions were motivated on the basis of morphological features. Bödiker (1698: 109) was the first author who argued that all of the six preterite present verbs, plus the former subjunctive of the past wollen, should be considered as auxiliaries, which is evidently motivated by the fact that all of these verbs can take infinitive complements. Yet, an explicit functional or grammatical motivation is missing for this classification. It was not until more than one hundred years later that Becker (1836: 176) explicitly provided such a motivation. As he argues, these verbs are particular in that they express either a possibility or a necessity. Accordingly, they should be considered as auxiliaries of mood. In his characterisation, the morphological aspect becomes less crucial, although Becker (1836: 176) stresses that these verbs are characterised by a morphological anomaly. As the class definition is primarily motivated in functional terms, he finishes by widening his scope and integrates two verbs into his class that behave analogously in semantic respect: lassen and brauchen. This line of reasoning has been adopted by a number of grammarians in the 19th century. Yet, most of them have tacitly excluded brauchen. However, their approach results in a class which is inconsistent in both respects, morphologically and functionally. Finally, Blatz (1896: 538) introduced the term Modalverba for reasons that remain mysterious. In the late 19th century, authors working in the tradition of the Neogrammarians, such as Braune (1886: 252), returned to a purely morphological definition. Accordingly, they dropped the functionally motivated designation ‘modal auxiliary’ and replaced it by a term that is solely based on morphological grounds: ‘praeterito praesentium’.

*Behaghel (1924) Basler (1935) Bech (1949) Dal (1951) Duden (1959) Griesbach/Schulz (1960) Welke (1965) Duden (1966) Duden (1973) Duden (1984) Duden (1995) Duden (1998) Duden (2005)

author Ölinger (1574) Claius (1578) Schottel (1663) Adelung (1782) Grimm (1822) Heyse (1822) Bauer H. (1832) Becker (1836) Bauer F. (1850) Bauer F./Duden (1887) Schoetensack (1856) Vernaleken (1861) Blatz (1896) *Braune (1886) Wilmanns (1906) Sanders (1908) Sütterlin (1910) Paul (1917) Curme (1922)

term (IPP) verba monosyllaba (IPP) unvollständige prädicate verba zweiter anomalie verba zweiter anomalie Huelfsverben des Modus Huelfsverben des Modus Hilfszeitwörter Hilfszeitwörter modale Huelfsverben verba zweiter anomalie Modalverba praeteritopraesentia Präterito Präsentien among Hilfsverben with IPP Präterito Präsentien Präterito Präsentien auxiliaries of mood, modal auxiliaries Hilfsverben Hilfszeitwort Modalverben modale Hilfsverben Modalverben Modalverben Modalverben Modalverben Modalverben Modalverben Modalverben Modalverben Modalverben x

x

x x x x

x x

x

x

taugen

thuerren

x x (x) x x

x

x

x (x)

wissen

x x x x x x x x x x x x x

x x x x x x x x x x x x x x x x x x

müssen

Tab. 2.6: Different extensions of the term Modalverb accros the centuries

x x x x x x x x x x x x x

können x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x

sollen x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x

mögen x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x

x x x x x x x x x x x x x x

dürfen x x x x

x x x x x x x x x x x x x

x

x

x x x x x x x x

wollen x x x x

x

x

x

x

x x x

lassen

x

(x)

x

x

brauchen

274 | 2 Modal verbs: A class struggle

2.3 The origin of the term Modalverb

|

275

In the late forties, Bech (1949) once again came up with a definition that is primarily functionally motivated. As his class only encompasses the traditional six modal verbs, other verbs with an analogous interpretation, such as brauchen and lassen, are ignored. Thus, from a strictly functional perspective, the choice made by Bech (1949) is not plausible. Furthermore, he is the one who re-introduced the term Modalverb, which was not much in use before and already forgotten. As can be seen, it is mainly due to the popular Duden grammars, and the grammar published by Griesbach and Schulz (1960: 65), that the term Modalverb in combination with the traditional six members could proliferate in the second half of the 20th century. As Bech (1949) is reluctant to reveal his references, it will remain a matter of speculation what authors were his major influences. Yet, it is an established fact that the term Modalverb had hardly been employed before him, except for Blatz (1896: 538) and Rossmann (1908). Moreover, the number of authors who suggested that the class encompasses exactly the traditional six verbs is fairly limited: before Bech, this extension was only promoted by few prominent authors, such as Bauer (1850: 102), Vernaleken (1861: 94) and Curme (1922: 317). In most descriptions until the thirties, lassen was considered as a modal auxiliary as well. With the functionally motivated definition proposed by Griesbach and Schulz (1960: 65), the availability of an epistemic interpretation became relevant as an essential characteristic of the modal verbs. In the subsequent period, the attention shifted to other verbs that exhibit an epistemic interpretation as well, such as werden, which was analysed as a modal verb by Welke (1965: 12) and Vater (1975: 110). The importance of this feature further increased until it became the core of the definition, e.g. in the accounts based on poly-functionality advocated by Öhlschläger (1989: 132), Engel (1996: 463), Diewald (1999: 1) and Reis (2001: 287). As this definition is rather functionally motivated than morphologically motivated, some of the authors adapt the extension of the class accordingly. As werden and brauchen exhibit an epistemic interpretation as well, Engel (1996: 463) and Reis (2001: 287) integrate these verbs into their respective classes of modal verbs. With the increasing attention that modal verbs received in the seventies and eighties, other items came into consideration that had not been part of the traditional class of modal verbs. Kolb (1964), Folsom (1968) and Scaffidi-Abbate (1973) illustrated that brauchen behaves like a modal verb in almost all of the essential aspects. Likewise, Öhlschläger (1989: 7) and Wurmbrand (2001: 137, 183, 224) demonstrate that möchte cannot be interpreted as subjunctive of the past of mögen; rather, it has to be treated as an independent lexical item. In the same spirit, Wurmbrand (2001: 137, 224) suggests that dürfte is a verb of its own type.

276 | 2 Modal verbs: A class struggle

In Section 2.2.13, it turned out that it appears impossible to establish a homogeneous class of modal verbs in German. The question thus arose to what extent these inconsistencies are due to the term and the concept Modalverb. After having reviewed their origin and development, it was seen that the history of the term, and the concept of Modalverb, is dominated by the conflict between a morphologically motivated definition and a functionally motivated definition. Starting with Becker (1836), most authors try to provide a definition that is motivated both morphologically and functionally. Yet, this approach does not succeed, as the class of verbs with the morphological anomaly is not co-extensive with the class of verbs that exhibit a functional peculiarity. Interestingly, most authors who provide a ‘double motivation’ for the class of modal verbs in German do not provide an explicit justification for this move. Thus, they fail to prove that each of the suggested verbs fulfils indeed both morphological and functional criteria. Any of the accounts mentioned above can be characterised as being either (rather) morphologically motivated, or (rather) functionally motivated. From this it follows that there are only two consistent ways of unifying the German verbs under consideration. Firstly, it is possible to provide a definition that is merely grounded on morphological aspects. Such a classification will result in a class equivalent to the preterite presents described by the Neogrammarian authors, such as Braune (1886: 252). Alternatively, one could provide a definition that only takes into consideration functional or grammatical aspects. Yet, as functional properties are often not too salient and their classification is often subject to controversies, the second approach faces even more challenges. As was revealed in Section 2.2.13, there is at least one functional criterion that is a fairly salient characteristic: The availability of an epistemic interpretation. Accordingly, the most fruitful solution appears to be to restrict the attention to the epistemic interpretations of all of the verbs considered here. As the resulting group is homogeneous in semantic respect, they could be regarded as a semantic or functional class. Even if the associated members do not constitute a morphological class, there are other formal properties they have in common: all of the epistemic modal verbs select bare infinitive complements, and all of them are raising verbs. In opposition to the modal auxiliaries of English, a definition that is motivated by both morphological and functional aspects will not result in a consistent class in German. Accordingly, it appears to be more appropriate to content ourselves with a class definition that is limited to the epistemic interpretations. At the same time, advocates of the more traditional definitions of modal verbs are confronted with the question of what benefit they would gain in cleaving to a class that is maybe elegant but not consistent and homogeneous. Yet, this does not mean that the anomalous morphology and the peculiar function do not have any impact on each other. As far as grammaticalisation is con-

2.3 The origin of the term Modalverb

|

277

cerned, it is clear that ‘anomalous’ forms are more likely to grammaticalise than ‘regular’ ones. A similar observation was made by Grimm (1822: 851), who noticed that verb forms that are frequent or exhibit an anomalous morphology act often as auxiliaries. Even if it is not possible to define a class of modal verbs in German, one can compare the kinship of the verbs under consideration. As it turned out, these verbs exhibit a couple of properties which are very rare, or at least not very frequent. For instance, a genuine epistemic interpretation is only attested with the verbs discussed here. Likewise, only the six traditional modal verbs, plus brauchen and lassen, exhibit an obligatory IPP-effect, as was illustrated in Section 2.1.1.2. Furthermore, the number of verbs that carry a preterite present or related morphology is fairly limited. This affects, in particular, the two first criteria: The 3rd person singular indicative without a suffix, and the vowel alternation between the singular present indicative forms and their plural counterparts. By contrast, the vowel alternation between the infinitive and the past tense stem is a characteristic that affects most of the irregular verbs as well. In more detail, Eisenberg et al. (2005: 491–502) lists more than 190 irregular verbs that display a vowel alternation between the present stem and the past stem. Apart from the verbs considered here, there are only a few subject-to-subject raising verbs in German, such as scheinen, drohen, versprechen, and pflegen. In a similar vein, the number of subject-to-object raising verbs is fairly limited (sehen, hören, fühlen, haben, etc.). As was mentioned in Section 2.1.2.1, example (20), there are only about ten more verbs in German that select bare infinitive complements, besides the potential modal verbs. Moreover, the number of predicates that select verbless directive phrases, other than the verbs investigated here, does not exceed ten, as was pointed out in Section 2.2.1.4. Thus, there are at least eight properties that cluster around the potential modal verbs and that are not regularly found with other verbs. These features could be helpful to determine the degree of kinship between the verbs listed in Table 2.7. Now, we could assume that features which are only observed with the potential modal verbs are more crucial for a definition. The column with the header “relevance” displays the number of verbs that exhibit the property apart from the verbs under consideration. Likewise, features that occur with only one or a few of the potential modal verbs would carry less weight. From a contemporary perspective, it might appear awkward that lassen was considered a modal auxiliary across centuries. Yet, it is more related to the traditional six modal verbs than it may seem at first glance. As has been demonstrated by Reis (2001: 308), lassen exhibits an epistemic interpretation, under certain conditions. Obviously, this use is restricted to environments in which lassen is part of a directive speech act.

278 | 2 Modal verbs: A class struggle

(511)

A: Was kostet das Buch wohl? what costs the book maybe

(512)

B: Lass es mal 100,– kosten. let-imp it part 100 cost-inf A: ‘How much could the book be?’ B: ‘Let it be 100,– /Assume that it costs 100,–.’

In the example given by Reis (2001), the epistemic use of lassen embeds a stative predicate, which is reminiscent of the other epistemic modal verbs considered in the previous section. Moreover, the speaker expects the embedded proposition not to be part of the deictic centre’s knowledge. In this respect, lassen behaves just as canonical epistemic modal verbs do. Interestingly, the deictic centre is not identical to the speaker here. Rather, it is linked to the addressee, who is grammatically encoded as the subject argument of lassen. A similar context shift can be observed with the reportative modal verbs wollen and sollen. Finally, it merits closer attention that lassen occurs more often than ordinary lexical main verbs and exhibits a frequency similar to other traditional modal verbs such as können, müssen or wollen, as was illustrated in Section 2.1.4. At this point, the striking analogies between lassen and the traditional six modal verbs cannot be denied anymore. Even if this approach would enable us to determine the degree of kinship between the potential modal verbs, it will hardly ever result in a clear-cut class of modal verbs. Accordingly, it is favourable to refrain from using the traditional concept of modal verbs. In this section, it has been shown that the traditional class of modal verbs is by far less natural than it appears. Moreover, this concept suffered from an essential conflict since its first introduction. It was claimed that it is motivated by morphological characteristics and functional characteristics. Yet, there has always been a discrepancy between these two classes of criteria. A definition that is based on the morphological anomaly will result in another class extension than a definition which is grounded on merely functional aspects. Furthermore, it has been illustrated that throughout the centuries there have been different terms and different extensions. Therefore, it should not be too surprising if the term Modalverb in combination with the six to eight verbs investigated here was not the last word in this debate.

relevance 0 0 0 0 reportative sollen > other epistemic modal verbs

How can the varying behaviour of these verbs be accounted for? Evidently, these differences are due to procedures of how the variable of the deictic centre is instantiated. As far reportative modal verbs are concerned, the variable of the deictic centre is anchored to an argument referent who is introduced by the modal verb itself: In the case of reportative wollen, the deictic centre is linked to the overt subject argument, in the case of reportative sollen, it is linked to a covert argument which is contextually identified. As can be seen, this process of identification is subject to further restrictions which could explain why reportative sollen is less readily acceptable compared to reportative wollen. The saliency of the deictic centre appears to be important at this point. Crucially, the variable is bound in a very local domain. Turning to epistemic modal verbs, the variable of the deictic centre may be linked to the speaker, to the addressee or to another referent. As can be seen, there are very strict conditions which have to be met in order to identify the deictic centre. In the most canonical case, the variable for the deictic centre introduced by the epistemic modal operator is bound over a large domain. This, in turn, means that the variable remains open for a long time. There are good reasons to assume that there are certain operators, such as circumstantial modal operators and certain tense operators, which do not tolerate free variables of this type in their domain. Accordingly, all verbs that introduce variables which are bound at a local level, such as reportative modal verbs, can occur embedded under such operators. By contrast, verbs that introduce variables that cannot be identified in the

5.2 So-called ‘evidentials’ drohen, versprechen and scheinen | 503

scope of that operator and that are otherwise left open are not acceptable in such contexts. Apart from the diagnostics discussed here, there are more statements about reportative modal verbs in non-canonical environments. On the one hand, some authors, e.g. Öhlschläger (1989: 236), emphasise the differences between reportative modal verbs and epistemic modal verbs. As he argues, reportative interpretations are more readily acceptable in event-related causal weil-clauses and in patterns in which they receive nuclear stress. On the other hand, there are authors, such as Ehrich (2001: 167), who focus on the environments from which reportative modal verbs are excluded, e.g. the scope of negation.

5.2 So-called ‘evidentials’ drohen, versprechen and scheinen As was shown in Sections 2.2.11 and 2.2.12, the raising verbs scheinen, dünken, as well as drohen, versprechen and verheißen differ from the traditional modal verbs in essential respects, regardless of what has been claimed in earlier literature. The empirical data provided in recent studies shows, in particular, that the hypothesis advocated by Askedal, (1997, 1998: 61) and Wurmbrand (2001: 205), according to which these raising verbs constitute a uniform class together with epistemic modal verbs, cannot be maintained. In the present Section, it will be demonstrated that the raising variants of these verbs differ significantly with respect to the non-canonical environments from which epistemic modal verbs are banned. In particular, these raising verbs are attested in contexts in which epistemic modal verbs have proven to be unacceptable: (i) they can be embedded as past participles under past tense auxiliaries; moreover, (ii) they are frequently attested in antecedents of event related conditionals; finally, (iii) there are contexts in which these raising verbs yield an interpretation which differs from the one of canonical epistemic modal verbs, e. g. in past tense contexts. In what follows, only the three frequent verbs scheinen, drohen and versprechen will be considered. Thus, the more archaic verbs dünken and verheißen will be ignored.

5.2.1 Past participles As was shown in Section 4.2, the use of epistemic modal verbs as past participles is fairly restricted. Basically, such uses can only be found embedded by present tense auxiliaries that are inflected for subjunctive of the past. In canonical present

504 | 5 Reportative and evidential modal operators

perfect tense or past perfect tense contexts, epistemic modal verbs cannot occur in German. Likewise, Askedal (1997b: 14), Fagan (2001: 220 Fn. 34) and Wurmbrand (2001: 205) claim that the raising uses of drohen and versprechen are banned from non-finite environments. As has been illustrated by Reis (2005: 133, 2007: 38) and Colomo (2011: 260– 265), the past participle of the raising pattern of drohen is well attested in German. The participle gedroht can be found in four different environments: (a) present perfect indicative (cf. 931–933), (b) present perfect subjunctive of the present (cf. 934), (c) past perfect indicative (cf. 935–941), and (d) past perfect subjunctive of the past (cf. 942). By contrast, epistemic modal verbs are only attested in environment (d), which represents the least important context for gedroht, which mainly occurs in past perfect contexts. (931)

Der zweitägige Volksentscheid in Rumänien über eine neue the two.day referendum in Romania about a new EU-gemäße Verfassung hat an einer zu geringen Beteiligung zu EU-conform constitution has at a to small participation to scheitern gedroht.²⁷ fail-inf threaten-ppp ‘The two-day referendum in Romania about a new EU-conform constitution was about to fail because of the too low participation.’

(932)

Wegen eines Lecks im Schiffsrumpf hat im due a leak in.the body has in.the Main-Donau-Kanal an der Schleuse Hilpoltstein Main.Donau-Channel at the water.gate Hilpoltstein (Landkreis Roth) ein Tankfrachter zu sinken administrative.disctrict Roth a tank.ship to sink gedroht.²⁸ threaten-ppp ‘Due to a leak in the body, a tank ship was about to sink in the Main-Donau-Channel at the water gate Hilpoltstein (administrative district Roth).’

(933)

Die spanischen Erpressungsversuche, die in letzter Minute the Spanish extortion.attempts rel.prn in the last den Beitritt zu verzögern gedroht haben, zeigen, minute the to procrastinate threaten-ppp have illustrates wie unfertig diese Union ist.²⁹ how incomplete this Union is

27 DeReKo: RHZ03/OKT.14345 Rhein-Zeitung, 20/10/2003. 28 DeReKo: NUN09/FEB.02429 Nürnberger Nachrichten, 21/02/2009.

5.2 So-called ‘evidentials’ drohen, versprechen and scheinen | 505

‘The Spanish attempts to extort which almost happened to delay the entry in the last minute almost illustrate how incomplete this Union is.’

(934)

Die Frau habe mit den Armen gezappelt und the woman has-sbjv.prs with the arms fidget-ppp and umzukippen gedroht.³⁰ to.fall-inf threaten-ppp ‘The woman fidgeted with the arms and was about to fall.’

(935)

und so schied ich mit günstigem Wind von dem Ufer, welches and so departed I with beneficial wind from the shore which mir lästrygonisch zu werden gedroht hatte.³¹ me laestrygonic to become threaten-ppp hatte ‘and so I departed from the shore that was about to become laestrygonic to me, supported by beneficial wind.’

(936)

Dabei hatte das Gewitter am späten Nachmittag die actually had the thunderstorm on.the late afternoon the Veranstaltung schon zu vereiteln gedroht.³² event already to thwart-inf threaten-ppp ‘Actually, the thunderstorm in the late afternoon was already about to thwart the event.’

(937)

Am 28. Mai hatten die Hochwasserdämme des an der on.the 28th May had the flood.dykes the-gen at the Ortschaft vorbeifließenden Mains nach heftigen Regenfällen locality bypassing Main-gen after strong rainfall zu brechen gedroht.³³ to burst-inf threaten-ppp ‘On 28th of May, the flood dykes of the Main that bypasses the locality were about to burst.’

29 DeReKo: P94/DEZ.42489 Die Presse, 24/12/1994. 30 DeReKo: BRZ08/DEZ.10451 Braunschweiger Zeitung, 19/12/1008. 31 DeReKo: GOE/AGI.00000 Goethe: Italienische Reise, [Autobiographie], (Geschr. 1813–1816), In: Goethes Werke, Bd. 11. – München, 1982 [p. 35]. 32 DeReKo: R97/JUL.54255 Frankfurter Rundschau, 15/07/1997. 33 DeReKo: NUN06/SEP.01329 Nürnberger Nachrichten, 12/09/2006.

506 | 5 Reportative and evidential modal operators

(938)

Somit endete ein Spiel für die Berner in euphorischem Jubel, so ended a game for the Bernese in euphoric exultation das zu einem weiteren Ärgernis der noch jungen Saison zu that to a further annoyance the-gen still young saison to werden gedroht hatte.³⁴ become-inf threaten-ppp had ‘And so ended the game for the Bernese with euphoric exultation that was about to become another annoyance in this still very young season.’

(939)

Nachdem er unter Martin Andermatt zu versauern gedroht after he under Martin Andermatt to shrivel-inf threaten-ppp hatte, blühte er zuletzt unter dem neuen Chef Petkovic auf.³⁵ had blossomed he recently under the new boss Petkovic out ‘After he was about to waste away under Martin Andermatt, he recently blossomed under the new boss Petkovic.’

(940)

Direktor Karl-Heinz Waibel erinnerte an die von Raiffeisen director Karl-Heinz Waibel reminded at the from.the Raiffeisen spontan bereitgestellte 1 Mill. S zur Bewerbung spontaneously provided 1 million Shilling to advertisement des Frühjahrsskilaufs, nachdem Schnee- und the-gen spring.skiing-gen after snow and Lawinenchaos die Gäste abzuschrecken gedroht avalanche.chaos the visitors to.off.frighten-inf threaten-ppp hatten.³⁶ had ‘Director Karl-Heinz Waibel reminded everyone of the one million Shilling provided by Raiffeisen for the promotion of the spring skiing project after the snow and avalanche chaos was about to frighten off the visitors.’

(941)

Das Bauvorhaben hatte sogar komplett zu platzen the building.project had even completely to fail-inf gedroht.³⁷ threaten-ppp ‘The building project was even about to entirely fail.’

34 35 36 37

DeReKo:SOZ07/JUL.04839 Die Südostschweiz, 26/07/2007. DeReKo: SOZ08/OKT.00350 Die Südostschweiz, 02/10/2008. DeReKo: V99/MAI.22846 Vorarlberger Nachrichten, 12/05/1999. DeReKo: NUZ09/SEP.02880 Nürnberger Zeitung, 30/09/2009.

5.2 So-called ‘evidentials’ drohen, versprechen and scheinen | 507

(942)

Die Schließung sei notwendig geworden, weil es the closing-down is-sbjv.prs necessary become-ppp because it aggressive, lautstarke und mit Alkohol verbundene Aktivitäten aggressive, loud and with alcohol combined activity auswärtiger Jugendlicher gegeben habe, die from.outside-gen adolescents-gen give-ppp have-sbjv.prs rel.prn aus dem Ruder zu laufen gedroht hätten.³⁸ out the rule to run-inf threaten-ppp have-sjbv.pst ‘Reportedly, the closing down became necessary because of aggressive, loud activities in combination with alcohol caused by adolescents from outside that were about to get out of control otherwise.’

There are a couple of facts that deserve closer attention. Firstly, the great majority of the occurrences involve the past perfect tense. Secondly, the infinitive complement is very often realised by the verb werden ‘become’. Thirdly, most of the examples which have been found in the corpus study stem from Switzerland, Austria or Southern Germany. As was shown in Section 2.2.12, the raising pattern of versprechen occurs significantly less frequently than the one of drohen. Moreover, it is almost restricted to infinitive complements that are realised by werden. Yet, there are a few instances of past participle versprochen used as a raising verb, as is illustrated in examples (943)–(944). Once again, the preferred configuration is the past perfect tense. (943)

Was ein sehenswertes Derby der Handball-Landesliga what a worth.seeing derby the-gen hand.ball.regional.league zu werden versprochen hatte, verkam in den Augen manches to become-inf promise-ppp had became in the eyes some-gen Zeugen zur Lachnummer.³⁹ witness to.a laughingstock ‘What promised to become an exciting derby in the regional league of handball turned into a laughingstock for many witnesses.’

(944)

Was am Samstag noch ein veritables Verkehrschaos zu what on Saturday still a true traffic.chaos to versprochen hatte und als solches auch vermeldet werden become-inf promise-ppp had and as such also announced worden war, stellte sich Sonntag und Montag als halb so pass.aux was put on Sunday and Monday as half so schlimm dar.⁴⁰ bad there

38 DeReKo: BRZ08/FEB.13559 Braunschweiger Zeitung, 26/02/2008. 39 DeReKo: RHZ00/NOV.20239 Rhein-Zeitung, 28/11/2000.

508 | 5 Reportative and evidential modal operators

‘What had promised to become a real traffic chaos on Saturday and what was also announced as such, turned out to be not as bad on Sunday and Monday.’

As Reis (2005: 133, 2007: 38) observes, the past participle is possible with the raising verbs drohen and versprechen, but it is not attested with the raising verb scheinen. As she argues, none of the three forms come into consideration: neither the regular form of the past participle gescheint, nor the irregular one geschienen, nor the IPP pattern scheinen. Even if Reis’ claims are well supported for Contemporary German, it merits closer attention that the raising verb scheinen could be used as a past participle in earlier stages of German. As illustrated below, the participle geschienen was in use as a raising verb in the 18th and 19th century. This form occurs in different configurations: In perfect tense indicative (cf. 951), in perfect tense embedded under subjunctive of the present (cf. 945), in past perfect tense indicative (cf. 950). Apart from that, Lessing frequently uses geschienen with ellipsis of the tense auxiliary (cf. 947–949). Of course, in these examples it is questionable to what extent geschienen can still be considered as a genuine past participle. (945)

Voltaire sagte, Frélon werde in der englischen Voltaire said Frélon pass.aux.sjbv.prs in the English Urschrift am Ende bestraft; aber so verdient diese Bestrafung original at.the end punished but so deserved this punishment sei, so habe sie ihm doch dem Hauptinteresse zu be-sbjv.prs so have-sbjv.prs she him yet the main.interest to schaden geschienen; er habe sie also weggelassen.⁴¹ harm-inf seem-ppp he have-sbjv.prs she thus omitted ‘Voltaire said Frélon was punished in the end in the English original. However, even if this punishment was justified, it seemed to have harmed the main interest. Thus, he omitted it.’

(946)

Seitdem die Neuberin, sub auspiciis Sr. Magnifizenz des since the Neuberin sub auspiciis his magnificence the-gen Herrn Prof. Gottscheds, den Harlekin öffentlich von ihrem Sir Prof. Gottsched-gen the harlequin publicly from her Theater verbannte, haben alle deutsche Bühnen, denen daran theatre banned have all German stages rel.prn there.on gelegen war, regelmäßig zu heißen, dieser Verbannung lied was regularly to be.called-inf this ban beizutreten geschienen. Ich sage, geschienen; denn im to.join-inf seem-ppp I say seemed because in.the

40 DeReKo: P92/JAN.00445 Die Presse, 07/01/1992. 41 Gotthold Ephraim Lessing, Hamburgische Dramaturgie, I, p. 96, (1767).

5.2 So-called ‘evidentials’ drohen, versprechen and scheinen | 509

Grunde hatten sie nur das bunte Jäckchen und den Namen reason had they only the colourful jacket and the name abgeschafft, aber den Narren behalten.⁴² abolish-ppp but the fool keep-ppp ‘Since the Neuberin sub auspiciis of His Magnificence Sir Prof. Gottsched has publicly banned the harlequin from her theatre, all of the German stages which wanted to be considered as conforming with the rules seemed to have joined this ban. I say ‘seemed’ because basically they have only abolished the colourful jacket and the name, but they have kept the fool.’

(947)

er muß sich gedulden, bis es der Ausgang lehre, he must refl be.patient-inf until it the outcome teach-sbjv.prs daß er da seiner Königin am getreuesten gewesen that he there his Queen at.the faithful-sup be-ppp sei, als er es am wenigsten zu sein geschienen.⁴³ be-sbjv.prs when he it at.the least to be-inf seem-ppp ‘He must be patient until it becomes clear that he was most faithful to his Queen when he seemed to be it the least.’

(948)

Bis auf den Augenblick, da er den Antenor ersticht, nimmt until of the instant where he the Antenor stabs takes auf die entschlossenste er an den Verbrechen seines Herrn he at the crimes his-gen master-gen at the resolute-sup Weise teil; und wenn er einmal Reue zu empfinden manner part and when he once remorse to feel-inf geschienen, so hatte er sie doch sogleich wieder seem-ppp so had he she yet immediately again unterdrückt.⁴⁴ suppressed ‘Until the moment when he stabs Antenor, he resolutely participates in the crimes of his master and, when he seemed to feel remorse once in a while, he had immediately suppressed it.’

(949)

Wenn wir also die Schönheiten dieser Figur durch und if we so the beauties this-gen figure through and durch untersuchen, so werden wir mit Grunde urteilen, daß through investigate so will we with reason judge that das, was man bisher für unbeschreiblich vortrefflich an ihrem this what one up.to.now for indescribably excellent at their

42 Gotthold Ephraim Lessing, Hamburgische Dramaturgie, I, p. 138, (1768). 43 Gotthold Ephraim Lessing, Hamburgische Dramaturgie, II, p.74, (1768). 44 Gotthold Ephraim Lessing, Hamburgische Dramaturgie, I, p.148, (1767).

510 | 5 Reportative and evidential modal operators

allgemeinen Anblicke gehalten, von dem hergerühret hat, was general view considered from this arose has what ein Fehler in einem Teile derselben zu sein geschienen.⁴⁵ a mistake in a part the.same-gen to be-inf seem-ppp ‘If we investigate the beauty of this figure thoroughly, we will conclude that what was considered as indescribably excellent arose from that, which seemed to be a flaw in the part of this figure.’

(950)

Schon in den Grundzügen der romantischen Erfindung already in the foundations the-gen romantic-gen concept erkannte ich den Dichter nicht wieder, der bis dahin allen recognised I the poet neg more the until then all Dingen eine erheiternde Seite abzusehen gewußt, mit dem things a exhilarating side to.off.see-inf know-ppp with the Mysticismus des Christentums sich nie befaßt, mysticism the-gen Christianity-gen refl never occupy-ppp überhaupt zur religiösen Poesie weder Anlage noch Neigung even to.the religious poetry neither talent nor affinity zu haben geschienen hatte.⁴⁶ to have-inf seem-ppp had ‘Even in the foundations of the romantic concept I could not recognise the poet, who until then managed to see an exhilarating side in all things, who never occupied himself with the mysticism of Christianity and who seemed to have no talent for or affinity to religious poetry.’

(951)

Ich hoffe nicht, daß Fritsche aus seiner sehr knauserigen I hope neg that Fritsche of his very parsimonious Oekonomie auch diesen zurükbehalten hat. Hat er das, so habe economy also this kept had had he this so have ich freilich bisher Unrecht zu haben geschienen⁴⁷ I of.course so.far wrong to have-inf seem-ppp ‘I do not hope that Fritsche kept it due to his parsimonious economy. If he had so, I seemed to have been wrong, of course.’

45 Gotthold Ephraim Lessing, Laokoon: oder über die Grenzen der Mahlerey und Poesie, p. 230, (1766). 46 Ernst Schulze, Cäcilie, I, preface, p. XII (1818). 47 Moritz Weinhold, Achtundvierzig Briefe von Johann Gottlieb Fichte und seinen Verwandten, p. 19, (1862).

5.2 So-called ‘evidentials’ drohen, versprechen and scheinen |

511

5.2.2 Event related conditional clauses As was shown in Section 4.17, epistemic modal verbs are subject to restrictions with respect to their acceptability in antecedents of event related conditional clauses. This behaviour is obviously due to the necessity to identify the variable for the deictic centre within a given context. The identification of the free variable underlies a couple of strict conditions. Accordingly, the sentence will be ungrammatical if these conditions are not fulfilled and the variable remains unbound. In contrast to epistemic modal verbs, the raising verb drohen can be used in antecedents of event related conditional clauses without any restrictions, as illustrated in examples (952)–(954) below. In all of these examples, it is possible to substitute the subordinator wenn by falls. This ensures that the wenn-clauses under investigation are really event related conditional clauses, rather than generic temporal clauses. (952)

Darin wird auch festgehalten, was passiert, wenn die Firma in.this is also recorded what happens if the company ihre Ziele nicht erreicht oder das Unternehmen sogar zu scheitern her goals neg reaches or the company even to fail-inf droht.⁴⁸ threatens ‘In this document, it is also specified what will happen if the company does not accomplish its goals or if the company showed indications of failing.’

(953)

Unterstützung erhielt der Nachwuchs von erfahrenen support got the offspring by experienced Feuerwehrmännern, aber nur dann, wenn etwas schief zu fire.men but only then if something bad to laufen drohte, wie etwa bei der Fahrzeugtechnik.⁴⁹ go-inf threatened as for.instance at the automotive.engineering ‘The offspring was supported by experienced fire men only if something showed indications of turning bad, as, for instance, regarding the automotive engineering.’

(954)

Wenn jemand plötzlich im Wasser Probleme bekommt und if somebody suddenly in.the water problems gets and zu ertrinken droht, kommen sie und helfen.⁵⁰ to drown-inf threatens come they and help ‘If somebody runs into problems while being in the water and threatened with drowning, they will come and help.’

48 DeReKo: M07/OKT.05535 Mannheimer Morgen, 24/10/2007. 49 DeReKo: RHZ06/OKT.28361 Rhein-Zeitung, 30/10/2006. 50 DeReKo: BRZ09/OKT.08437 Braunschweiger Zeitung, 19/10/2009.

512 | 5 Reportative and evidential modal operators

Moreover, the raising verb drohen abundantly occurs in generic temporal wennclause, which is another environment that is hardly compatible with epistemic modal verbs. Likewise, raising patterns of versprechen are attested in the antecedent of event related conditionals. Again, the conjunction wenn can neatly be replaced with falls, which clearly indicates that the examples below are really event related conditional clauses, rather than generic temporal wenn-clauses. (955)

Man werde nur dann zukaufen, wenn dies profitabel zu one will-sbjv.prs only then back.buy if this lucrative to verspreche.⁵¹ werden become-inf promise-sbjv.prs ‘As is said, they will only buy it back if this business promises to be lucrative.’

(956)

Wo man singt, da lass Dich nieder, sagt ein Sprichwort, und where one sings there let you down says a saying and diesem folgt denn auch jung und alt gerne; zumal this follows then also young and old willingly especially dann, wenn die Nächte sommerlich warm zu werden then when the nights summerly warm to becomeversprechen.⁵² promise ‘As the saying goes, “where they sing, there you shall settle down” which old and young people obey willingly, specially if the nights promise to become summerly warm.’

Finally, the raising patterns of scheinen can also be found in antecedents of event related conditional clauses, as is exemplified in examples (957)–(958). (957)

Wenn er Fähigkeiten zu haben scheint, die man sich nicht if he abilities to have seems rel.prn one refl neg erklären kann, dann nennen wir ihn einen Zauberer oder explain can then call we him a sorcerer or Magier.⁵³ magician ‘If he seems to have abilities that cannot be explained, we call him a sorcerer or magician.’

51 DeReKo: HAZ08/MAR.02647 Hannoversche Allgemeine, 13/03/2008. 52 DeReKo: A98/JUN.36812 St. Galler Tagblatt, 04/06/1998. 53 DeReKo: BRZ07/NOV.01811 Braunschweiger Zeitung, 15/11/2007.

5.2 So-called ‘evidentials’ drohen, versprechen and scheinen |

(958)

513

Wenn es einem aber vor allem darum zu gehen scheint, if it one yet above all about.that to go-inf seem fragwürdige Entscheidungen der SPD-Landesregierung questionable decisions the-gen SPD.regional.government zu rechtfertigen, dann schreibt man in seinem Leserbrief to justify then writes one in his letter.to.the.editor natürlich etwas anderes.⁵⁴ of.course something else ‘If it appears to be only about justifying questionable decisions made by the regional government led by the SPD, than you will write something else in your letter to the editor.’

Aside from event related conditional clauses, the three raising verbs drohen, versprechen and scheinen are also abundantly recorded in generic temporal wennclauses, which is another related environment from which epistemic modal verbs are excluded. These facts clearly indicate that these raising verbs significantly differ from epistemic modal verbs. At this point the question arises why these raising verbs are acceptable in event related conditional clauses whereas epistemic modal verbs are not. As was illustrated above, epistemic modal verbs introduce a variable for the deictic centre which has to be anchored to an appropriate attitude holder. This process of identification underlies strict conditions. If these conditions are not fulfilled, the variable remains unbound and the linguistic structure cannot be interpreted. As was shown, the antecedents of event related conditional clauses are environments in which these conditions are difficult to meet for epistemic modal verbs. Evidently, drohen, versprechen and scheinen differ with respect to the nature of the deictic centre and how it is identified. Turning to drohen and versprechen, Reis (2005: 140, 2007: 18) has pointed out a whole range of essential analogies which they share with aspectual verbs. Accordingly, she suggests considering the two raising verbs as aspectual verbs. In Section 2.2.12, more arguments in favour of this type of analysis have been presented. If drohen and versprechen are considered as aspectual verbs, the necessity to assume a deictic centre decreases. Assuming that these lack such a deictic centre, there is no variable which needs to be bound and, correspondingly, they are not subject to the conditions of variable binding. As far as scheinen is concerned, the situation is different. As was demonstrated in Section 2.2.11, scheinen can optionally realise its deictic centre as a dative NP. Accordingly, it is plausible to assume that the deictic centre is always

54 DeReKo: RHZ09/OKT.09156 Rhein-Zeitung, 10/10/2009.

514 | 5 Reportative and evidential modal operators

represented as an argument of the raising verb, even if it is not overtly realised. As can be seen, this argument position is usually filled with a generic pronoun, which refers to a contextually given group of persons or the totality of all human beings. Apart from that, any analysis which treats scheinen as an epistemic modal verb ignores one important detail: It can alternatively select hypothetical comparative clause. Thus, it appears to be more appropriate to consider scheinen as a verb that compares two state of affairs. The one which the speaker is considering, and the one to which it resembles. In other words: the state of affairs under consideration exhibits the same characteristics as the characteristics of the state of affairs expressed by the embedded proposition.

5.2.3 Past tense As was observed by Reis (2005: 129, 2007: 13) and Colomo (2011: 241–245), the raising pattern of drohen and versprechen behave in a very different manner compared to epistemic modal verbs when they are inflected for the past tense. As Reis argues, the past tense of the raising verb drohen does not convey ‘a report of speaker inferences on the basis of present knowledge but an objective report of a past event’. This indicates that drohen and versprechen are not evaluated to the same extent with respect to the speaker’s knowledge as epistemic modal verbs are. In a similar manner, scheinen is attested in past tense context in which it does not undergo a context shift which is typical of epistemic modal verbs in such environments, as demonstrated at length in Section 4.3. (959)

Mozart schien das Leben in Italien zu genießen.⁵⁵ Mozart seem-pst the life in Italy to enjoy-inf ‘Mozart seemed to enjoy his life in Italy.’

(960)

Mozart [...] schien sich in Mailand so wohl zu fühlen, daß er Mozart seem-pst refl in Milan so well to feel-inf that he seiner Schwester mitteilte, daß er ‘keine lust mehr auf his sister told that he no desire anymore on salzburg habe’.⁵⁶ Salzburg have-sbjv.prs ‘Mozart seemed to feel so well in Milan that he told his sister that he did not feel like going back to Salzburg.’

55 Kurt Palm, Der Mozart ist fett und wohlauf, Wien: Löcker, p. 66, (2005). 56 Kurt Palm, Der Mozart ist fett und wohlauf, Wien: Löcker, p. 111, (2005).

5.2 So-called ‘evidentials’ drohen, versprechen and scheinen |

(961)

515

Allerdings schien Mozarts Vater auch vom neuen Plan however seem-pst Mozart-gen father also by.the new plan nicht wirklich überzeugt zu sein.⁵⁷ neg truly convince-ppp to be-inf ‘However, Mozart’s father did not seem to be very convinced about the new plan.’

(962)

Berufliche Angelegenheiten schienen den Vater in dieser professional affairs seem-pst the father in this Situation freilich nicht zu interessieren.⁵⁸ situation of.course neg to interest-inf ‘Of course, the father did not seem to be interested in business matters in this situation.’

(963)

Je länger Leopolds Aufenthalt in Wien dauerte, desto mehr the longer Leopold-gen stay in Vienna lasted the more Gefallen schien er an diesem Leben zu finden.⁵⁹ pleasure seem-pst he an that live to find-inf ‘The longer Leopold’s stay in Vienna continued, the more he seemed to like this sort of life.’

The examples in (959)–(963) are particularly revealing regarding the nature of the evaluation at work in the case of scheinen because they involve an author who talks about a period in which he did not live yet. Accordingly, the past tense form schien in example (959) does not reflect an assumption of the author which has been made when Mozart was in Italy. As a consequence, it does not yield a past speech act event reading or a reported indirect speech interpretation, which is the natural interpretation for epistemic modal verbs in this context. Alternatively, one could assume that schien reflects an assumption of Mozart’s contemporaries. But this type of interpretation is not the adequate one either. At this point, the question arises what the precise contribution of the past tense marker is in these contexts. It appears to be similar to the past event reading, an assumption made by the speaker at the Time of Utterance based on some evidence from the past. It is not clear what this type of analysis can look like in more detail. Based on the hypothesis that scheinen is a verb which compares two states of affairs, an attractive solution can be found. If the main contribution of scheinen is the expression of a comparison between two states of affairs, it is expected that the past tense operator will indicate that one of the two states of affairs ceased to exist. Accordingly, example (959) could be rephrased in the following way: ‘There

57 Kurt Palm, Der Mozart ist fett und wohlauf, Wien: Löcker, p. 118, (2005). 58 Kurt Palm, Der Mozart ist fett und wohlauf, Wien: Löcker, p. 218, (2005). 59 Kurt Palm, Der Mozart ist fett und wohlauf, Wien: Löcker, p. 274, (2005).

516 | 5 Reportative and evidential modal operators

was a state of affairs which was similar to the hypothetical state of affairs in which Mozart liked life in Italy’. Moreover, the referent who can perceive the comparison can optionally be encoded by a dative NP. In German, it is possible to use scheinen with respect to past events or states in two different ways. The past can be expressed as a past tense morpheme on the verb scheinen, as is illustrated above. Moreover, the infinitive complement can be realised as a perfect tense infinitive, as illustrated below. Although the meanings of these two alternatives seem to overlap, they are by no means synonymous. As demonstrated in example (965), there are even contexts in which the past tense form schien cannot be substituted with a present tense form scheint, which embeds a perfect tense infinitive. Interestingly, a replacement is significantly more acceptable with epistemic modal verbs which are evaluated at the Time of Utterance (cf. 966–968): (964)

Mozart scheint das Leben in Italien genossen zu haben. Mozart seem the live in Italy enjoy-ppp to have-inf ‘Mozart seems to have enjoyed the life in Italy.’

(965)

# Je länger Leopolds

Aufenthalt in Wien dauerte, desto mehr the longer Leopold-gen stay in Vienna lasted the more Gefallen scheint er an diesem Leben gefunden zu haben. pleasure seem he an that live find-ppp to have-inf

‘The longer Leopold’s stay continued, the more he seemed to like this sort of life.’

(966)

Je länger Leopolds Aufenthalt in Wien dauerte, desto mehr the longer Leopold-gen stay in Vienna lasted the more Gefallen dürfte er an diesem Leben gefunden haben. pleasure might he an that live find-ppp have-inf ‘The longer Leopold’s stay continued, the more he seemed to have liked this sort of life.’

(967)

Je länger Leopolds Aufenthalt in Wien dauerte, desto mehr the longer Leopold-gen stay in Vienna lasted the more Gefallen könnte er an diesem Leben gefunden haben. pleasure could he an that live find-ppp have-inf ‘The longer Leopold’s stay continued, the more he could have liked this sort of life.’

(968)

? Je länger Leopolds

Aufenthalt in Wien dauerte, desto mehr the longer Leopold-gen stay in Vienna lasted the more Gefallen muss er an diesem Leben gefunden haben. pleasure must he an that live find-ppp have-inf

‘The longer Leopold’s stay continued, the more he must have liked this sort of life.’

These contrasts follow naturally if scheinen is analysed as a verb that establishes a comparison between a given state of affairs and a hypothetical state of affairs.

5.2 So-called ‘evidentials’ drohen, versprechen and scheinen |

517

5.2.4 Summary As this section has briefly demonstrated, the raising verbs scheinen, drohen and versprechen behave differently from epistemic modal verbs. There are a couple of environments in which they occur and in which epistemic modal verbs are categorically ruled out, e.g. past participles which are embedded by indicative perfect tense auxiliaries and event related conditional clauses. Aside from that, there are environments in which these raising verbs yield interpretations in which they differ from genuine epistemic modal verbs, such as in past tense contexts. From this it follows that drohen, versprechen and scheinen cannot be analysed as epistemic modal verbs, as claimed by Askedal (1997: 14, 1998: 61) and Wurmbrand (2001: 205). Alternatively, there are good arguments to treat drohen and versprechen as aspectual verbs, as has been suggested by Reis (2005: 140, 2007: 18). By contrast, scheinen is most efficiently captured as a verb that establishes a comparison between a given state of affairs and a hypothetical state of affairs. The referents who can perceive this comparison can optionally be realised as a dative NP. As a consequence, scheinen, drohen and versprechen will not receive any further consideration in the upcoming sections.

6 Anchoring the deictic centre In the preceding sections, it was shown that epistemic modal verbs are excluded from a whole range of environments: (i) they do not occur with verbless directional phrase complements, (ii) they cannot be separated from their infinitive complements in wh-clefts, (iii) they do not undergo nominalisation, (iv) they are banned from adverbial infinitives, and finally, they cannot be embedded under (v) circumstantial modal verbs, (vi) predicates of desire, (vii) imperative operators, or (viii) optative operators. By contrast, reportative modal verbs are attested in some of these environments: in nominalisations (iii), in adverbial infinitives (iv), embedded under tense auxiliaries (v), and optative operators (viii). Moreover, it was demonstrated that there are environments in which epistemic modal verbs and reportative modal verbs obtain different interpretations. In most of these environments, reportative modal verbs do not undergo context shift, e.g. in past tense contexts, in questions and in antecedents of event related conditionals. The aim of this chapter is to provide an explanation for the following observations. 1. Why are epistemic modal verbs excluded from these environments? 2. Why are reportative modal verbs nevertheless possible in certain of these environments? 3. Why do the two types of modal verbs differ in their interpretations in certain contexts? 4. Why do reportative wollen and reportative sollen have different preferences for these environments? As it turns out, epistemic and reportative modal verbs can be characterised as operators which introduce a variable for the deictic centre. In order to apply the Condition for Deictic Centres (CoDeC), this variable needs to be instantiated. This operation of identification needs to fulfil certain anchoring conditions. If these conditions are not met, the variable remains unbound. As can be seen, there are particular contexts that do not tolerate an unbound variable of the deictic centre.

6.1 The speaker, the addressee and arguments Obviously, epistemic modal verbs are evaluated with respect to a certain attitude holder. In the most frequent case, this is the speaker. Yet, there are contexts in which the epistemic modal verb is evaluated with respect to a referent other than

https://doi.org/10.1515/9783110411027-006

6.1 The speaker, the addressee and arguments |

519

the speaker. Accordingly, the epistemic operator must be some sort of variable which identifies the attitude holder who makes a commitment to a certain believe.

6.1.1 Declarative speech acts In their most frequent use, epistemic modal verbs are part of a declarative speech act. There is a broad consensus that the epistemic modal verb is evaluated with respect to the speaker’s knowledge in these configurations. (969)

So kann die Motte in Wipshausen einmal ausgesehen haben.¹ so can the Motte in Wipshausen once out.look-inf have-inf ‘The Motte in Wipshausen may once have looked like this.’

(970)

Die Kleidungsstücke deuten dann auch darauf hin, dass es sich the clothes indicate then also to.it at that it refl um einen Mann gehandelt haben müsste.² about a man deal-ppp have-inf must-sbjv.pst ‘The clothes indicate that it must have been a man.’

(971)

Der Wunsch nach Ungestörtheit dürfte schließlich dem Liebespaar the wish for privacy might finally the love.couple auf so tragische Weise das Leben gekostet haben.³ for such tragic manner the live cost-ppp have-inf ‘The wish for privacy might have finally caused the death of the lovers, who died in such a tragic manner.’

In all of the representative examples given above, the attitude holder who makes the epistemic assumption is identified with the speaker. Thus, the speaker indicates that the embedded proposition is not part of his own knowledge.

6.1.2 Interrogative speech acts Much more rarely, there are epistemic modal verbs that occur in information seeking questions. As was illustrated in great detail in Sections 4.11 and 4.12, the epistemic modal is not evaluated with respect to the speaker. Rather, the attitude holder who carries the belief is identified with the addressee.

1 DeReKo: BRZ09/AUG.04565 Braunschweiger Zeitung, 11/08/2009. 2 DeReKo: RHZ09/JUN.24827 Rhein-Zeitung, 29/06/2009. 3 DeReKo: NON09/JAN.04467 Niederösterreichische Nachrichten, 12/01/2009.

520 | 6 Anchoring the deictic centre

(972)

a.

MOPO: Wer könnte die Fälschung der Wahl veranlasst who could the fraud the-gen election arranged haben? have

b.

Steinbach: Nur einer: Revolutionsführer Ajatollah Ali only one revolution.leader Ajatollah Ali Chamenei.⁴ Chamenei ‘Who could have arranged the fraudulent elections? Only one, the leader of the revolution Ajatollah Ali Chamenei.’

A speaker who uses an epistemic modal verb in an information seeking question indicates that he does not expect the addressee to have enough knowledge to commit himself to a precise answer. Rather, the speaker asks the addressee about his assumptions. Accordingly, the speaker does not attribute the embedded proposition to the addressee’s knowledge. It was illustrated in Sections 4.11 and 4.12 that question operators are operators that induce a general type of context shift in which the addressee becomes the most salient referent.

6.1.3 Complement clauses As was shown in Section 4.15, the epistemic modal verb is not evaluated with respect to the speaker. Based on the observations made by Lasersohn (2005: 277), Stephenson (2007: 489) argues that an embedded epistemic modal verb is always evaluated with respect to an attitude holder argument of the immediately superordinate predicate. Likewise, Zimmermann (2004: 265) argues that related epistemic modifiers such as the discourse particle wohl may never take scope out of a complement clause. (973)

Und er erzählt, dass dieses Rätsel bald gelöst sein könnte.⁵ And he tells that this riddle soon solv-ppp be-inf could ‘And he said that this riddle could be solved soon.’

(974)

Polizeisprecher Thomas Figge erklärte gestern auf Anfrage, police.spokesman Thomas Figge declared yesterday on demand dass der 33-Jährige mindestens Tempo 100 gefahren sein muss.⁶ that the 33.year.old at.least tempo 100 drive-ppp be-inf must

4 DeReKo: HMP09/JUN.01442 Hamburger Morgenpost, 17/06/2009. 5 DeReKo: NUN03/AUG.02519 Nürnberger Nachrichten, 28/08/2003.

6.1 The speaker, the addressee and arguments |

521

‘The police spokesman Thomas Figge declared yesterday, on demand, that the 33year-old must have driven at least 100 km/h.’

In both examples, the epistemic modal verb is embedded under a predicate which involves a subject referent that can be interpreted as an attitude holder. Accordingly, the variable of the deictic centre provided by the epistemic modal verb is anchored to the subject referent er ‘he’ in example (973), and to the subject referent ‘police spokesman Thomas Figge’ in example (974). In the examples above, the speaker signals that he does not attribute the embedded proposition to the knowledge of the respective subject referents.

6.1.4 Reportative modal verbs As was demonstrated in Sections 2.2.3.7 and 2.2.6.4, reportative modal verbs attribute a claim to one of their arguments. In the case of wollen, the claim is associated with the subject argument referent (975) and, in the case of sollen, the claim is associated with a covert argument which is contextually identified (976). (975)

Badhapur ist ein Sadhu, ein Weiser, Gerechter. 106 Jahre will die Badhapur is a Sadhu, a sage righteous 106 years wants the hagere Gestalt mit dem langen grauen Haar schon alt sein.⁷ rawboned figure with the long grey hair already old be-inf ‘Badhapur is a Sadhu, a wise man, a religious man. This haggard form with long grey hair claims to be as many as 106 years old.’

(976)

Tom Cruise und Katie Holmes sind geschockt. L. R. Hubbard (kl. Tom Cruise and Katie Holmes are shocked L R Hubbard small F.) soll Suris Vater sein.⁸ picture shall Suri-gen father be-inf ‘Tom Cruise and Katie Holmes are shocked. L. R. Hubbard is claimed to be Suri’s father.’

In the examples above, there appear to be two referents that come into consideration as the deictic centre: Firstly, the deictic centre could be the speaker. Secondly, the deictic centre could be the attitude holder argument of wollen and sollen. As will be demonstrated in the next section, there are good reasons to adopt the latter point of view. According to this approach, the speaker indicates that he has

6 DeReKo: BRZ09/MAI.05146 Braunschweiger Zeitung, 12/05/2009. 7 DeReKo: NUN99/OKT.02110 Nürnberger Nachrichten, 23/10/1999. 8 DeReKo: HMP08/JAN.00616 Hamburger Morgenpost, 08/01/2008.

522 | 6 Anchoring the deictic centre

no evidence whether the embedded proposition is really part of the verified knowledge of the experiencer argument. In example (975), he would convey that he does not know whether the proposition Sadhu is 106 year old is really part of Sadhu’s verified knowledge. Alternatively, it could be a false belief or Sadhu could be lying. In neither of the two cases would the speaker consider the proposition as Sadhu’s knowledge. In example (976), the speaker signals that he has no reason to assume that the proposition L. R. Hubbard is Suri’s father is really part of the covert argument referent’s knowledge. This will be shown in great detail in Section 6.2.2.

6.1.5 Summary In this section, it was demonstrated that epistemic modal operators are not always evaluated with respect to the speaker. There are particular contexts and configurations in which the operator is interpreted with respect to other referents: In information seeking questions, the relevant referent is the addressee; in embedded context, the epistemic agent is realised as an attitude holder argument of the superordinate clause. Finally, the operator is evaluated with respect to an argument of the modal verb in the case of reportative modal verbs. These facts reveal the nature of the syntactic and semantic representation of that attitude holder. As it can refer to different referents, it has to be some sort of variable. In what follows, this variable will be referred to as the ‘deictic centre’. The following sections are dedicated to the conditions governing the instantiation of that variable.

6.2 The deictic centre As was illustrated above, epistemic modal operators introduce a variable with respect to which they need to be evaluated. A similar concept was introduced by Charles Fillmore (in the reprinted version: Fillmore (1997: 98)) in his Lectures on Deixis, in the early seventies, in which he discusses the deictic centre for the first time. Subsequently, it was developed in more detail by Levinson (1983: 64). A similar notion has already been developed by Bühler (1934: 102), who refers to it as Origio. Abraham (2011: xxxv) illustrates that the notion of Origo is a potential means for the description of epistemic modality. Even if the deictic centre is a concept that affects a lot of linguistic elements, this section will uniquely be addressed to its role for epistemic modal operators. There are further reasons that make it plausible to assume that epistemic modal verbs introduce a variable for the deictic centre. Firstly, there is an independent need to assume that any modal operator specifies a modal source. As

6.2 The deictic centre |

523

will be shown in Section 6.2.1, this modal source is identical to the deictic centre in the case of epistemic modal verbs. Secondly, it was illustrated throughout this study that epistemic modal verbs are characterised as operators that are evaluated with respect to someone’s knowledge. In particular, their use indicates that the speaker does not know whether the embedded proposition really holds. How this can be formulated in more precise terms will be pointed out in Section 6.2.2.

6.2.1 The modal source Necessities and possibilities are often considered as abstract forces. Accordingly, the modal source is the source of these forces. As for circumstantial modal verbs, this modal source is typically instantiated by the individual who imposes the obligation in the case of müssen ‘must’, who grants the permission in the case of dürfen, and to whom the volition is attributed in the case of wollen and sollen. An influential description of the modal source was provided by Bech (1949: 4). He argues that, sometimes, the modal source is already specified in the lexicon entry. As Bech (1949: 37) exemplifies, wollen and sollen need to be analysed as necessity modal verbs. Accordingly, wollen and sollen can be considered as necessity modal verbs which specify their modal source as [+internal]. In contrast, there are other necessity modal verbs such as müssen which remain underspecified with respect to the modal source. Diewald (1999: 102) has shown that the modal source is always realised by the speaker in the case of epistemic modal verbs. In other words, the speaker is the referent who judges the embedded proposition with respect to its validity and who makes a commitment to the truth. In contrast, the discourse referent who makes the commitment to the truth in the case of reportative modal verbs is the referent encoded by the experiencer argument, as was observed Diewald (1999: 225). Interestingly, the modal source has not received much attention in the most popular approaches in formal semantics such as Kratzer (1981), Kratzer (1991), Brennan (1993) or Hacquard (2006). In one of the most prominent approaches on modality, Kratzer (1991: 649) assumes modal operators involve three relevant dimensions of modality: the modal force, the modal base and the ordering source. Yet, the modal source does not appear to play any crucial role in this account. But in the past decades, there has been a growing number of studies which have illustrated the necessity of the concept modal source. In his formal semantic analysis, Westmoreland (1998: 74) illustrates that an epistemic operator always has to be evaluated with respect to a given person. Likewise, Abraham (2005: 263) argues that both circumstantial and epistemic modal verbs involve a modal source which is represented as a covert argument of the modal verb. In a similar fashion,

524 | 6 Anchoring the deictic centre

Depraetere and Verhulst (2008: 3) demonstrates that each type of necessity has its source of modality, including epistemic necessities. Furthermore, Lasersohn (2005) indicates that predicates of personal taste have to be evaluated with respect to a judge, which is most typically instantiated by the speaker. Extending his analysis to epistemic modal verbs, Stephenson (2007: 497) shows that epistemic modal verbs also have to be evaluated with respect to some judge.

6.2.2 The Condition on Deictic Centres Throughout this study, it was shown that the most efficient way to characterise epistemic modality is in terms of knowledge. A speaker who employs an epistemic modal verb indicates that he does not know whether the embedded proposition is true. This condition was referred to as the Condition on Deictic Centres (CoDeC). The next section is dedicated to the question of how it can be formulated in a precise way. The section after that will illustrate how the CoDeC can be applied to reportative modal verbs. 6.2.2.1 Which propositions are not part of the knowledge? In principle, there are three ways to formulate the CoDeC. The open question is which propositions exactly are not part of the deictic centre’s knowledge. Is it only the positive proposition (977a)? Or is it rather the negated proposition (977b)? Or is it the conjunction of them which has to be excluded from the deictic centre’s knowledge (977c)? (977) Three ways to formulate the CoDeC: a.

p is not part of the deictic centre’s knowledge

b.

¬p is not part of the deictic centre’s knowledge

c.

neither p nor ¬p part of the deictic centre’s knowledge

As can be seen, there are advocates for each of the three positions. As most of the authors are not very explicit about this matter, it is not always clear whether the classifications given below really reflect their intentions. Nevertheless, they can roughly be associated with one of the three options. The first position, according to which the positive proposition only is excluded, (977a) appears to be defended by Krämer (2005: 60, 133) and Ziegeler (2006: 90). A compatible account was suggested by Diewald (1999: 207), who claims that, in the case of epistemic modal verbs, the deictic centre values the embedded proposition as [± non-factive]. In a less explicit way, Erb (2001: 161), Fintel and Gillies (2010: 353), Kratzer (2011, 2012: 99) argue for an fairly similar analysis based on the concept of direct evidence rather

6.2 The deictic centre |

525

than knowledge. The second position, according to which the negated proposition only is excluded, (977b) is explicitly advocated by Martin (2011: Sect. 3.1.), who claims that a speaker who employs an epistemic modal verb ‘is not sure in EVAL-T that P is false’. Finally, the third position, according to which both propositions are excluded (977c), appears to be entertained by Westmoreland (1998: 12), though he does not make any explicit claim about this matter. As there are configurations in which the speaker knows the embedded proposition to be false, the alternatives (977b) and (977c) are less plausible. It was shown in great detail in Sections 2.2.1.5, 2.2.2.5, 2.2.6.5 and 4.4 that epistemic modal verbs can be combined with false proposition if they are inflected for the subjunctive of the past. In particular, this affects the three forms könnte, müsste and sollte. A representative example is given below (cf. 978): (978)

Wenn alle Meldungen über Schwangerschaften der if all reports about pregnancies the-gen Oscar-Preisträgerin gestimmt hätten, müsste sie oscar-winner attune-ppp have-sbjv.pst must-sbjv.pst she mittlerweile 30 Babys bekommen haben. Kidman ist Mutter meanwhile 30 babies get-ppp have-inf. Kidman is mother zweier adoptierter Kinder.⁹ two-gen adopted-gen children ‘If all of those reports about the Oscar winner’s pregnancies had been true, then she would have had 30 babies by now. Kidman is the mother of two adopted children.’

As this instance clearly shows, the author of that utterance knows that Nicole Kidman does not have 30 children. Of course, one could argue that this peculiar behaviour is the effect of the subjunctive of the past form. But an appropriate analysis remains to be developed. The second position, according to which only the negated proposition is excluded, (977b) is the least plausible one. Apart from the difficulties just mentioned, it cannot account for the fact that an epistemic modal verb hardly ever embeds a proposition which is known to be true. The first option has the great advantage of capturing subjunctive of the past modals as well without any further stipulations. And as will be shown in the following sections, it is able to capture the behaviour of reportative modal verbs as well. At this point, it is possible to formulate the Condition for Deictic Centres as follows: (979) Condition on Deictic Centres (CoDeC)

9 DeReKo: BRZ07/DEZ.11819 Braunschweiger Zeitung, 31/12/2007.

526 | 6 Anchoring the deictic centre

The use of an epistemic operator indicates that the embedded proposition is not part of the deictic centre’s knowledge.

This conditions neatly captures the behaviour of epistemic modal verbs in declarative clauses, in questions and in embedded complement clauses. In the following section, it will be demonstrated that it is also capable of accounting for the behaviour of reportative modal verbs. 6.2.2.2 The deictic centre in reportative modal verbs As was indicated above, reportative modal verbs crucially differ from epistemic modal verbs with respect to the deictic centre because they involve two different potential candidates which come into consideration for the deictic centre: the speaker and the referent encoded by the experiencer argument provided by the reportative modal verb. By means of the diagnostics presented in the previous section, it will become more evident which of the two candidates acts as the deictic centre for reportative modal verbs. Once again, the relevant aspect concerns the knowledge of the referents. Granted that the CoDeC as is formulated above holds, two hypotheses will be examined: (i) the use of a reportative modal verb implies that p is not part of the speaker’s knowledge, and (ii) the use of a reportative modal verb implies that p is not part of the experiencer’s knowledge. Hypothesis (i) is refuted in case reportative modal verbs occur in contexts in which p is part of the speaker’s knowledge, hypothesis (ii) is refuted in case reportative modal verbs occur in contexts in which p is part of the knowledge attributed to the referent encoded by the experiencer argument. First of all, Ehrich (2001: 157), Colomo (2011: 241), Faller (2011: 4) and Faller (2012: 289) have pointed out that reportative modal verbs can embed propositions which the speaker knows to be false. Such configurations are attested in corpora, as is illustrated in examples (980)–(982). In this respect, reportative modal verbs differ from epistemic modal verbs that are inflected for the indicative. (980)

Die Familie des angeblichen Verlobten weiß nichts von the family the-gen alleged fiancé knows nothing about einer Tania Head. Die Bank Merrill Lynch, bei der sie a Tania Head the bank Merrill Lynch at rel.prn she gearbeitet haben will, hatte sie nie auf der Gehaltsliste.¹⁰ work-ppp have-inf wants had her never on the payroll

6.2 The deictic centre |

527

‘The family of her alleged fiancé has never heard about Tania Head. At Merrill Lynch, where she claims to have worked, she was never on the payroll.’

(981)

Es war nicht korrekt, diesen Druck auf den angeklagten it was neg correct that pressure on the accused Kindermörder auszuüben. Aber dass sich dieser Strolch vor dem child.murderer to.exert-inf but that refl that thug of the Polizei-Vize-Chef „gefürchtet” haben soll, ist Schauspielerei.¹¹ police-vice-boss afraid.be-ppp have-inf shall is comedy ‘It was not correct to exert pressure on the accused child murderer. But he is a comedian if the thug claims to have been “afraid” of the vice-boss of the police.’

(982)

bei mir in der Firma soll angeblich ein Paket bei mir at my in the company shall reportedly a parcel at me abgegeben worden sein.... Stimmt nicht. Ich habe das Paket deliver-ppp pass.aux-ppp be-inf holds neg I have the parcel nie gesehen.¹² never seen ‘Reportedly, a parcel was delivered to me at my company .... Not true. I have never seen the parcel.’

These instances reflect situations in which the speaker knows the embedded proposition to be false and in which the knowledge of the referent expressed by the experiencer argument is left unspecified. This referent could know the embedded proposition to be false, thus he would be lying. Alternatively, he could also have a false belief and be convinced that the embedded proposition is true. As a consequence, the effect of the reportative modal verb is to label the commitment of that referent as unreliable. In a similar manner, Diewald (1999: 228) suggests that reportative wollen occurs even more often in environments in which the speaker raises doubts about the validity of the embedded proposition. An analogous reasoning is advocated by Öhlschläger (1989: 235). Even if examples (980)–(982) are in slight favour of hypothesis (ii), according to which the relevant knowledge is the one associated with the experiencer argument, they do not refute hypothesis (i). Hypothesis (i) is only refuted if there are contexts in which the embedded proposition is really part of the speaker’s knowledge. As illustrated by the discourse

10 As quoted in Colomo (2011: 241): DeReKo: HAZ07/OKT.00069 Hannoversche Allgemeine, 01/10/2007. 11 As quoted in Colomo (2011: 241): DeReKo: RHZ04/DEZ.17444 Rhein-Zeitung, 18/12/2004. 12 As quoted in Faller (2012: 289): Urbia.de forum post, http://www.urbia.de/archiv/forum/th3614401/Unterschrift-gefaelscht-Paket-verschwunden-Und-nun.html, last accessed 25/7/2012.

528 | 6 Anchoring the deictic centre

given in example (983), such cases exist. Assume that the speaker is a doctor who talks about a hypochondriac patient. (983)

a.

Der Schani ist ein alter Hypochonder, andauernd kommt er the Schani is a old hypochonder always comes he mit anderen Geschichten daher. with other stories along

b.

Stell Dir vor, jetzt willquot er Malaria haben. imagine you part now want he malaria have-inf

c.

Und soll ich dir was sagen? Er hat wirklich Malaria, And shall I you something say he has indeed Malaria ich habe gerade die Blutproben vom Labor I have just the blood.test from.the laboratory zurückbekommen. back.get-ppp ‘Schani is an old hypochondriac. Each time he comes to see me, he tells a different story. Imagine, he now claims to have malaria. Believe it or not, I’ve just got back his blood test results from the laboratory and it says that he indeed has malaria.’

This example describes a context in which the referent encoded by the experiencer argument makes a non-verified claim about himself. Crucially, the proposition I have Malaria is not part of his knowledge, either he does not know whether it holds or he has a false belief. In contrast, the speaker knows that this referent indeed has Malaria. Arguably, the context given in example (983) could alternatively be interpreted in a way that the referent encoded by the experiencer argument already knew beforehand that he had malaria because he had already made another blood test at another hospital. This seems to refute hypothesis (ii). However, this context deserves a closer look. Even if it turns out that the referent encoded by the experiencer argument indeed already knew that he had malaria, the use of the reportative modal verb in example (983) contributes some resonance of doubt. How can this be accounted for? It is important to distinguish between what this referent really knows and what knowledge the speaker attributes to that referent. As it turns out, the latter type of knowledge is only relevant to the interpretation of reportative modal operators. In employing a reportative modal verb, the speaker expresses that he has no compelling evidence that the modified proposition p is indeed part of the referent’s knowledge. In some cases, he could have even explicitly known that p is not part of the referent’s knowledge. However, up to now, there are no examples of reportative modal operators attested in which the speaker knows that the modified proposition is true and part of the knowledge that is attributed to the referent encoded by the experiencer argument. Moreover, in this type of

6.2 The deictic centre |

529

context, a reportative modal operator would be redundant. As it seems, then, a speaker who employs a reportative modal verb expresses that he does not attribute the modified proposition p to the knowledge of the referent encoded by the experiencer argument. Reportative modal operators reflect the perspective of the speaker, rather than the objective truth. This explains why the speaker considers the information conveyed by reportatively modified propositions as not reliable even in contexts in which it later turns out that the referent encoded by the experiencer argument did indeed know that the proposition was true. Accordingly, the acceptability of reportative modal operators does not hinge on whether the referent encoded by the experiencer argument knows that he has malaria. Rather, the use of a reportative modal operator signals that the speaker does not attribute the modified proposition to the knowledge of the referent encoded by the experiencer argument. Faller (2012: 289) provides an authentic example taken from the web. Just like with example (983), example (984) exhibits a configuration in which the speaker knows the proposition to be true. Regarding the knowledge of the referent expressed by the experiencer, it is left unspecified: even if the option in which the uncle is lying is not very likely, it is nevertheless not excluded. In any case, the speaker conveys a glimpse of doubt whether the uncle’s conclusion is well grounded on reliable evidence or if it is true by chance. (984)

Die Standzeit soll sehr hoch sein laut Onkels Aussage the service.life shall very high be-inf according Uncle’s report und die Äste werden überhaupt nicht gequetscht, habe ich and the branches are at.all neg crushed have I auch selbst gesehen.¹³ also myself seen ‘The service life is, according to Uncle’s report, very high and the branches are not crushed at all, I have also seen it myself.’

However, it is necessary to remark that the validity of Faller’s example is not beyond doubt. The phrase habe ich auch selbst gesehen could also refer to the second conjunct (die Äste werden überhaupt nicht gequetscht), which does not contain any reportative modal verb. According to this configuration, the embedded proposition would not automatically be knowledge of the speaker. Yet, the interpretation suggested by Faller is possible, even if it was not intended by the speaker. Examples like (983) and (984) demonstrate two things: first, reportative modal verbs are compatible with situations in which the speaker knows that

13 As quoted in Faller (2012: 289): Werkzeug-News.de forum post, http://www.werkzeugnews.de/Forum/viewtopic.php?p=147095, last accessed 25/7/2012.

530 | 6 Anchoring the deictic centre

the embedded proposition is true. In such a context, the speaker signals that he does not know whether the proposition is really also part of the referent’s knowledge or whether this referent lies, has a false belief or utters a proposition without having any evidence for its truth. Secondly, the speaker does not ascribe the embedded proposition to the knowledge of the referent encoded by the experiencer argument. As a consequence, these observations provide evidence for hypothesis (ii). Thus, reportative modal verbs differ from their epistemic relatives in two important respects. While in the case of reportative modal verbs, it is the referent encoded by the experiencer argument which is committed to the truth of the embedded proposition, it is the speaker in the case of epistemic modal verbs. Furthermore, the use of an epistemic modal verb signals that the embedded proposition is not part of the speaker’s knowledge. By contrast, the use of reportative modal verbs indicates that the embedded proposition is not part of the knowledge ascribed to the referent expressed by the experiencer argument. Even if the CoDeC can be applied to the reportative modal verbs, this does not entail that it makes the right prediction for all modifiers that are related to epistemic modal verbs which were discussed in Chapter 3: As it turns out, they fail to capture the behaviour of relevance conditionals. In conclusion, the CoDeC can also be applied to reportative modal verbs. In this case, the deictic centre is instantiated by the referent expressed by the experiencer argument. Accordingly, reportative modal verbs are evaluated with respect to the knowledge of the referent encoded by the experiencer argument which is realised as the subject in the case of wollen, and as some other covert argument in the case of sollen.

6.2.3 A subtle refinement The attentive reader may raise an objection against the CoDeC as formulated in Section 6.2.2: the CoDeC appears to be formulated too strictly, as it suggests that the speaker has no knowledge about the embedded proposition at all – but this is certainly not the case. As with the utterance in example (985), the speaker already knows that the man has died. Likewise, the mother in example (986) knows that something bad happened to her daughter’s mouth. (985)

Der Mann dürfte im Schlaf gestorben sein, da die Beamten the man might in.the sleep die-ppp be-inf as the officers ihn im Bett gefunden hatten.¹⁴ him in.the bed found had

6.2 The deictic centre |

531

‘The man has probably died while sleeping, as the officers found him in his bed.’

Das Mädchen hatte im Garten gespielt und plötzlich Blut the girl had in.the garden played an suddenly blood gespuckt. Die Angst der Mutter: Das Kleinkind könnte spewed the fear the-gen mother the toddler can-sbjv.pst Glas verschluckt haben.¹⁵ glass swallow-ppp have-inf

(986)

‘The girl was playing in the garden and suddenly she started spewing blood. The mother was afraid the toddler could have swallowed glass.’

The answer to this question leads us to an interesting relation between epistemic operators and focus background structure, which could finally reveal the precise location of the epistemic operator with respect to the proposition. Back to example (985). The basic facts are (i) that the speaker knows the background and (ii) that the speaker knows a set of alternatives, which may be an open set, in which not all of the alternatives are known to the speaker. In contrast to a canonical assertion, the speaker is not in a position to commit himself to one of these alternatives. Formally, this could be modeled by assuming that in most of the epistemic worlds which are accessible to the speaker the man died while he was sleeping. However, in some of the worlds the man was about to get up, in others he took an overdose, in very few worlds he died in the kitchen and was put in there to his bed by his friend. In an account building on the tradition of Kratzer (1981) and Kratzer (1991), truth conditions would be sufficient in order to get the correct interpretation of her quantifiers which operate on sets of possible worlds. Maybe, an account in which the worlds are additionally weighted with respect to their probability, as suggested by Lassiter (2011: 163–164) would gain even better results for deontic and volitional modal operators. There are a couple of obvious alternatives as to which constituent is focused. All of them contribute different conditions of usage. If the speaker knows that the event under discussion took place during the sleep, uttering example (987a) is deceiving; however, the alternatives (987b) and (987c) are pragmatically felicitous in this context. Likewise, if the speaker knows that the subject referent died, example (987b) is deceiving and (987a) as well as (987c) are pragmatically felicitous, and so forth. (987)

a.

Der Mann dürfte im SCHLAFfoc gestorben sein. alternatives={während dem Aufstehen, beim Bett machen, in der Küche, . . . }

14 DeReKo: BVZ09/OKT.01155 Burgenländische Volkszeitung, 14/10/2009. 15 DeReKo: BRZ09/AUG.03524 Braunschweiger Zeitung, 08/08/2009.

532 | 6 Anchoring the deictic centre

b.

Der Mann dürfte im Schlaf geSTORBenF sein. alternatives={entführt worden, erschreckt, erkrankt, . . . }

c.

Der MANNF dürfte im Schlaf gestorben sein. alternatives={die Frau, der Hund, der Hamster, . . . }

As it turns out, the epistemic use of dürfte is focus sensitive to some moderate extent. This is reminiscent of the findings put forth by Jacobs (1988: 94), who found out that sentential adverbs can be sensitive to focus. Likewise, Bildhauer (2011: 364) points out, following Jacobs (1986), that these adverbs often occur fronted with focused constituents in the Forefield position, which highlights the particular relation between these two elements. In a similar fashion, Lassiter (2011: 113– 119) notices that the epistemic adjectives likely and probable are pragmatically sensitive to focus alternatives. A related observation was made by McDowell (1987: 236). She demonstrates that earlier claims are false, according to which epistemic utterances are always the result of the combination of a ‘known’ subject referent and an ‘unknown’ predicate. As she argues, the sentence (988) can be the result of three different epistemic conclusions. (988)

Olaf must have murdered Yuri. a.

Someone murdered Yuri and Olaf is most probably that someone.

b.

Olaf murdered someone and Yuri is most probably that someone.

c.

Olaf had something to do with Yuri’s death and murder is most probably what it was.

Another counterexample in support of McDowell’s analysis was given above in (985), in which the ‘unknown constituent’ is represented by the temporal adverbial im Schlaf. At this position, the contrast to assertions becomes very evident. Following the relational concept of the focus background structure advocated by Jacobs (1988: 93), the focus background structure is organised by illocutionary operators, most notably by the assertion operator ASS. As Jacobs (1988: 95) points out, the felicitous use of the ASS operator requires that conditions are met: (i) the speaker believes that the background is true, (ii) the speaker believes that the predications consisting of background and focus is true, (iii) the addressee believes that the background is true, (iv) the addressee does not know whether the predication holding between background and focus is true. Epistemically modified utterances do not fit neatly in this frame, as the speaker does NOT know whether the predication actually holds or not, conflicting with condition (ii). Rather, he considers it possible/likely or certain that this

6.3 A unified analysis for epistemic and reportative modality |

533

predication holds. The distinct behavior of epistemic utterances could indicate that epistemic operators actually operate on the level of illocution, in other words: they are illocutionary operators. Under such a view, epistemic operators would be defined in analogy to Jacobs’ (1988) ASS operator, with a modified condition (ii). Such an approach is reminiscent of Lyons’ (1977: 805) reasoning, who argues, based on intuitive grounds, that epistemic operators occupy the same structural position as question operators; a detailed discussion is given in Section 4.22.1. Yet, as it was shown in Section 4.23, there a couple of obstacles for this endeavour, as epistemic operators appear to be part of the proposition. Moreover, they occur embedded under information seeking polarity and wh-questions, which demonstrates that these two types of operators occupy different positions, as it was shown in the Sections 4.11 and 4.12. Hence, epistemic operators can only be illocutionary operators if there is a structural hierarchy of illocutionary operators. Furthermore, it is not clear how reportative modal operators could be integrated into this approach. Finally, there is another challenge that remains to be mastered. If epistemic operators really operate on the level of illocutions, it needs to be shown which contribution they make to Stalnaker’s (1978) common ground: As they do not provide new shared knowledge, their effect cannot be the elimination of those possible worlds which are not in line with the newly added proposition. However, all of these issues will be left for future research.

6.3 A unified analysis for epistemic and reportative modality As was seen in the previous section, there are different referents with respect to which an epistemic operator can be evaluated. At this point, it still remains mysterious what precisely governs the identification of the deictic centre. In the upcoming section, it will be demonstrated how deictic centres are anchored to appropriate referents. The identification is based on a hierarchy of salience, which ranks the most likely candidates that come into consideration for the deictic centre, as will be demonstrated in Section 6.3.1. This hierarchy turns out to be a powerful tool, as it is not only capable of explaining the correct identification of the deictic centre, it also accounts for the fact that epistemic modal verbs are excluded from the non-canonical environments, and that reportative modal verbs can occur in some of these environments at the same time, as will be pointed out in Section 6.3.2.

534 | 6 Anchoring the deictic centre

6.3.1 Hierarchy of Salience As it seems, the deictic centre of an epistemic operator is always anchored to the closest syntactically represented referent which can be interpreted as an attitude holder, whereas closeness is defined in terms of the syntactic clause hierarchy. Accordingly, the closest candidates are appropriate arguments which are introduced by the modal operator itself. Such configurations only exist with reportative modal verbs: Whereas wollen involves an experiencer argument which is realised as a subject, sollen involves a covert experiencer argument which is contextually supplied. The next candidates are appropriate arguments which are introduced by a predicate that embeds an epistemic modal verb. In the most typical case, these verbs are attitude predicates or predicates of communication. Finally, the last candidate is the most salient referent of the speech act, which is the speaker in declarative clauses and the addressee in information seeking questions and conditionals. As there is always such a referent, epistemic modal verbs are by default evaluated with respect to the most salient referent of the speech act unless there is another appropriate candidate which occurs in a closer distance to the epistemic operator. (989) Hierarchy of Salience 1. the experiencer arguments of the predicate which introduce the epistemic modal operator 2.

the experiencer argument of an attitude predicate in the superordinate clause

3.

the most salient referent of the speech act

Regardless of which of these candidates the epistemic modal operator will be anchored to, the CoDeC has to be applied. From this analysis it follows that despite their substantial differences reportative operators could be seen as epistemic operators which are already anchored at the level of the verb itself. Given these conditions of anchoring, it is finally possible to account for the incompatibility of epistemic modal verbs with the non-canonical environments discussed above.

6.3.2 Operators which impose selectional restrictions As was illustrated in Chapter 4, epistemic modal verbs are excluded from eight environments: (i) they do not occur with verbless directional phrase complements, (ii) they cannot be separated from their infinitive complements in wh-clefts, (iii) they do not undergo nominalisation, (iv) they are banned from adverbial infin-

6.3 A unified analysis for epistemic and reportative modality |

535

itives, and, they cannot be (v) embedded under circumstantial modal and other auxiliary verbs, (vi) predicates of desire, (vii) imperative operators or (viii) optative operators. Regarding these contexts, the question arises what these configuration have in common. Upon closer inspection, it turns out that the environments (iv)–(viii) involve some sort of circumstantial modal operator. As was pointed out by Nissenbaum (2005: 145), Grosz (2014), rationale clauses, which are closely related to the adverbial infinitives under discussion here involve a covert (circumstantial) modal operator. Furthermore, this circumstantial modal operator bears scope over the epistemic modal operator. As observed by many authors, such as Cinque (1999: 87), Eide (2005: 9) and Colomo (2011: 111), the scope of a circumstantial modal operator is an environment from which epistemic modal operators are excluded. Based on the findings made in the previous sections, it is now possible to provide an explanation for this behaviour. Evidently, circumstantial operators are a type of operator which cannot embed unbound variables for the deictic centre: (990)

Circumstantial modal operators fail to embed structures which contain an unbound variable for the deictic centre.

This condition makes a whole range of predictions. Among others, circumstantial modal operators should embed epistemic operators whose variable for the deictic centre is bound. In general, it is expected that configurations in which a circumstantial operator directly embeds an epistemic modal operator whose deictic centre is not bound (991) should be by far less acceptable than configurations in which a circumstantial operator embeds some sort of intervening attitude predicate under which the epistemic operator is embedded and which introduces a referent argument eligble to serve as attitude holder (992). (991) (992)

* CIRC [ EPIST(deictic centre=?) ] CIRC [ ATT(attitude holder) [ EPIST(deictic centre=attitude holder) ] ]

First of all, consider epistemic modal verbs which occur in the antecedent of a conditional. As shown in Section 4.17, the acceptability of epistemic modal verbs in antecedents of event related conditionals is fairly limited. If the condition stated under (990) is correct, it would be expected that the acceptability of epistemic modal verbs in antecedents of event related conditionals should increase if the modal is embedded by an intervening attitude predicate which introduces an argument that can bind the deictic centre introduced by the modal. The example below given by Fintel and Gillies (2008: 93), contains an antecedent in which the epistemic modal verbs is embedded by the attitude predicate realize:

536 | 6 Anchoring the deictic centre

(993)

If Blofeld realizes you might be in Zürich, you can breathe easy – he’ll send his henchman to Zürich to find you.

As shown in Section 4.17, it is not certain whether conditionals of the type illustrated in example (993) are indeed to be seen as event related conditionals. This issue is not really relevant here. What is crucial is the fact that the epistemic modal occurs embedded under an attitude predicate makes the whole conditional more acceptable. Secondly, if this condition is right, it is also expected that reportative modal verbs should be significantly more acceptable embedded under circumstantial modal operators. As was seen in Section 5.1, they are attested in adverbial infinitives, embedded under auxiliaries, and in optatives. The analysis proposed above gains further support by the behaviour of epistemic lassen, which was discovered by Reis (2001: 308). Interestingly, this pattern appears to be restricted to imperatives. As was demonstrated in Sections 2.1.1.3 and 4.13, the traditional six modal verbs lack an imperative form, except for wollen, which can rarely be found in such patterns if it is used without an infinitive complement. According to the selectional restrictions of circumstantial modal operators (cf. 990), it is expected that any epistemic operator that involves a bound variable for the deictic centre should in principle be acceptable in imperatives. As lassen has an independent imperative form and modal semantics, it provides an interesting case. In the case of epistemic lassen, the deictic centre is identified with the subject referent. As it is used as an imperative, the subject remains syntactically unrealised and it is identified with the addressee. If the CoDeC is correct, it is expected that the speaker does not attribute the embedded proposition it costs 100,–. to the knowledge of the addressee. And indeed, this appears to be the case in example (995). (994)

A: Was kostet das Buch wohl? what costs the book maybe

(995)

B: Lass es mal 100,– kosten. let-imp it part 100 cost-inf A: ‘How much could the book be?’ B: ‘Let it be 100,– /Assume that it costs 100,–.’

As predicted, the speaker expects that the embedded proposition is not part of the deictic centre’s knowledge. In this respect, lassen behaves just as canonical epistemic modal verbs do. In contrast, the deictic centre is realised as the subject argument of the epistemic modal verb in the case of lassen. As a consequence, the variable for the deictic centre is already instantiated within the scope of the imper-

6.4 Alternative analyses |

537

ative operator. As predicted by the analysis above, this results in a grammatical configuration. In a similar vein, nominalisation could be regarded as an operation which only applies to verbs that contain no unbound variable for the deictic centre. This would neatly account for the fact that reportative modal verbs can be subject to nominalisation whereas epistemic modal verbs cannot. There are only two of the non-canonical environments discussed in Chapter 4 which are not accounted for with respect to selectional restrictions of a superordinate operator: (i) the fact that epistemic modal verbs cannot be separated from their infinitive complements in wh-clefts and (ii) their incompatibility with verbless directional complements. Neither of the cases can be accounted for in terms of anchoring of the deictic centre. Interestingly, these environments are equally unacceptable for reportative modal verbs. As it seems, the reason to their ungrammaticality is due to the selectional restrictions of the epistemic modal verb itself. As was shown in Section 5.1.5, it appears to be possible under certain conditions to identify unbound variables of the deictic centre by a rule of accommodation, as proposed by Lewis (1979: 172) and Kratzer (1981: 61).

6.4 Alternative analyses There are other analyses which explain the restricted compatibility of epistemic modal verbs with the environments discussed in Chapter 4. The most explicit accounts for German were developed by Wurmbrand (2001: 182–204) and Erb (2001: 116–125), who argue that epistemic modal verbs have to be considered as auxiliaries which have lost all properties typical of the category ‘verb’. They exhibit an impoverished morphology and they cannot be used as non-finite forms (infinitive, past participle) anymore. Both approaches follow Cinque’s (1999) cartographic tradition, assuming that the different types of modal verbs are merged as different functional categories in the clausal hierarchy. In Wurmbrand’s (2001: 183) analysis, epistemic modal verbs are merged in Aux0 , circumstantial modal verbs with raising patterns in Mod0 and circumstantial modal verbs with control patterns in ʋ0 , as is illustrated on the left-hand side of Figure 6.1. Turning to Erb’s (2001: 124) approach, she suggests that epistemic modal verbs are merged in a functional category for sentence mood M0 , circumstantial modal verbs with raising structure in Mod0 and circumstantial modal verbs with control structure in V0 , as shown on the right-hand side of Figure 6.1. In both analyses, modal verbs with control syntax are considered as lexical verbs, as they involve proper referential subject arguments.

538 | 6 Anchoring the deictic centre AuxP

MP M

Aux′′ Aux0

ModP Mod ʋP

T

epistemic Mod0

ʋ0

epistemic T0

ModP/AspP

ʋ circumstantial VP

M0

TP

Mod /Asp Mod0 /Asp0

VP V

dynamic NP/VP/CP

semi-lexical (raising) V0 lexical (control)

Fig. 6.1: Wurmbrand (2001) (left) and Erb (2001) (right)

In essence, Wurmbrand and Erb argue that epistemic modal verbs are functional elements rather than lexical verbs and accordingly they have lost the ability to occur in non-finite contexts. Askedal (1997: 13, 1998: 60) has adopted a similar view. Analogous approaches have been suggested for English, cf. Butler (2003) and Roberts (2003). Unfortunately, these approaches face a whole range of serious challenges. Firstly, it was demonstrated in Section 4.1 and Section 4.2 that epistemic modal verbs can occur in non-finite environments under certain conditions: if the matrix predicate involves an argument which can be interpreted as an attitude holder. The existence of non-finite occurrences of epistemic modal verbs cannot be explained under the assumption that epistemic modal verbs are not verbs but affixlike entities which are merged in functional projections for inflection or more abstract elements.

6.4 Alternative analyses |

539

Secondly, it is not evident how these accounts treat reportative modal verbs. In particular, it is necessary to account for the parallel behaviour of reportative modal verbs and epistemic modal verbs. This concerns, on the one hand, their ability to embed predications consisting of an identified individual and a predicate which refers to a state that is not likely to change, or a predicate that refers to a past event. And on the other hand, their requirement to be anchored to some deictic centre. Given these properties, one could assume that reportative modal verbs are merged as a functional category which occupies a high position in the clausal hierarchy such as Aux0 or M0 . Yet, reportative modal verbs involve referential arguments: wollen selects an animate referential subject argument and sollen has an argument which is usually not overtly realised. Following the canonical assumption, higher functional projections do not contribute proper arguments. Accordingly, reportative modal verbs cannot be merged in the same functional projection as epistemic modal verbs. Alternatively, one could assume that reportative modal verbs are lexical categories and merged as V0 or ʋ0 . In this case, the necessity of reportative modal verbs to be anchored to a deictic centre appears to be detached from the status of their category. Accordingly, variables for deictic centres could also be introduced by lexical categories. As a consequence, the question arises why epistemic modal verbs need to be functional categories at all. Thirdly, it is not evident how these accounts could capture the fact that reportative modal verbs can be subject to nominalisation whereas their epistemic cognates cannot. Moreover, these accounts even fail to explain why circumstantial modal verbs with raising structure can be nominalised, as nominalisation is traditionally considered as an operation which only applies to lexical categories but not to functional ones. Fourthly, Wurmbrand’s (2001: 183) and Erb’s (2001) analyses cannot account for differences between reportative and epistemic modal verbs regarding nonfinite environments. Whereas epistemic modal verbs are very restricted in this respect, reportative modal verbs can occur in a couple of contexts from which their epistemic counterpart are excluded, such as adverbial ohne-zu infinitives or embedded under auxiliaries. In general, it is not clear to what extent these accounts are capable of capturing the differences between epistemic modal verbs and reportative modal verbs, as described in Chapter 4 and Chapter 5. Moreover, cartographic approaches, such as those presented here, are confronted with ordering paradoxes, as there are many more alternatives of combining modal verbs than these approaches would predict. This is discussed by Maché (2012: 132) in more detail. Considering these challenges, it appears to be more efficient to analyse all of the different syntactic patters of modal verbs as elements of the category ‘verb’. Reportative modal verbs and epistemic verbs are special in that they introduce a

540 | 6 Anchoring the deictic centre

variable for a deictic centre which needs to be bound. Both types of verbs only differ with respect to the domain in which the deictic centre is actually instantiated.

6.5 Summary In this chapter, it was pointed out that the distributions of epistemic modal verbs and reportative modal verbs can be accounted for in terms of anchoring of the deictic centre. Epistemic modal operators introduce a variable for the deictic centre with respect to which they are evaluated. In order to be interpreted, this variable needs to be instantiated by an appropriate attitude holder. The instantiation of the deictic centre follows the Hierarchy of Salience, according to which the variable will be anchored to the closest appropriate argument which can be interpreted as an attitude holder. In the most canonical case, the variable will only be identified at the level of the speech act, which results in a configuration in which the deictic centre is anchored to the speaker. Moreover, there are various contexts that do not tolerate unidentified variables. Circumstantial modal operators fail to embed linguistic structures which contain a free variable for a deictic centre. This explains the behaviour of epistemic modal verbs, which are generally banned from the scope of a circumstantial modal operator. Furthermore, it was shown that reportative modal operators can be considered as particular epistemic modal operators, which introduce a variable for the deictic centre that is anchored to an argument of the modal operator itself. As a consequence, these variables introduced by reportative modal verbs are already instantiated at a very local level. In turn, this accounts for the fact that reportative modal verbs can occur in the scope of circumstantial modal operators as they do not contain any unbound variable. As was pointed out, nominalisations are another environment which is not compatible with unbound variables for the deictic centre. As a consequence, they do not apply to epistemic modal verbs whereas they do apply to reportative modal verbs. The behaviour of circumstantial modal operators and nominalisations could be captured in terms of selectional restrictions: These operators are restricted to linguistic structures which do not contain a free variable for the deictic centre. From the facts discussed above it follows that, despite their substantial differences, reportative operators could be seen as epistemic operators that are already anchored at the level of the verb itself. Thus, neither epistemic modal verbs nor reportative modal verbs need to be regarded as functional categories. They can be neatly described as lexical verbs which are characterised by the fact that they introduce a variable for the deictic centre that needs to be bound according to a couple of conditions.

7 On black magic: A diachronic explanation Each syntactic pattern described in Section 2.2 corresponds to a different stage of grammaticalisation. Accordingly, the descriptions given in that section roughly trace the historical development of each individual verb. The grammaticalisation of epistemic modal verbs typically involves the steps indicated in (996). Similar trajectories were proposed by Lehmann (1995: 33) and Diewald (1999: 2, 34). (996) transitive verb ⇒ control verb with event modification ⇒ raising verb with event modification ⇒ epistemic verb. For most of the verbs considered above, the path of development is slightly different or even more complex. An extensive description of individual developments was given in Section 2.2.9, where brauchen is discussed, the youngest epistemic modal verb, which only grammaticalised in the 19th century. The following section will only address the last step of grammaticalisation, when circumstantial modifiers turn into epistemic ones, as we saw in the previous chapters that the epistemic patterns are the most essential ones for the verbs presently under investigation. As for the step from transitive verbs to verbs which select infinitive complement, the reader is referred to Paul (1920: 95) and Fritz (1997: 68), who argue that the bare infinitive complements have their origin in former accusative NPs. As has often been observed, it is fairly intricate to determine the precise interpretation of a given traditional modal verb. By means of the diagnostics developed in Chapters 2 to 6, the characteristics of circumstantial modal verbs and their epistemic modal counterparts were identified. Moreover, it was shown in which respects these two types of verbs differ. These differences are essential for an understanding of the diachronic development of modal verbs. Following the most prominent hypothesis advocated by Traugott (1989: 35), Sweetser (1990), Bybee, Perkins and Pagliuca (1994: 195), Fritz (1997), Diewald (1999), Axel (2001: 45), epistemic modal verbs emerged diachronically from their circumstantial cognates. In Section 7.1, a small corpus study on the behaviour of epistemic modal verbs from the 16th century will be presented. As was argued by Fritz (1991: 45, 1997: 94) and Diewald (1999: 365), this is the crucial period in which the use of epistemic modal verbs became frequent in German. As it turns out, almost all of the epistemic modal verbs found in this corpus select stative predicates that refer to states that cannot be changed. In contrast, circumstantial modal verbs typically subcategorise for infinitive complements that contain eventive predicates. Accordingly, there are good reasons to assume that the first epistemic modal verbs in history selected stative complements. A similar observation has already

https://doi.org/10.1515/9783110411027-007

542 | 7 On black magic: A diachronic explanation

been made by Abraham (1991), Abraham (2001) and Abraham (2005) and Leiss (2002). In Section 7.2, an account will be provided which explains why the first epistemic modal verbs in history selected stative predicates. This approach is based on a pragmatic rule of accommodation in the spirit of Lewis (1979: 172), which is considered as black magic by Kratzer (1981: 61).

7.1 Epistemic modal verbs in Early New High German As is known, polyfunctional modal verbs occurred in Germanic languages from the early Middle Ages on. Krause (1997: 95) discusses a whole range of potentially epistemic modal verbs from Old High German. In most of the cases, the relevant verb is mugan. Yet, as Axel (2001: 45 Fn. 31) has pointed out, the status of many of her examples is rather doubtful. At any rate, there are a few examples of mugan which are epistemic beyond doubt, e.g. example (997). Likewise, Denison (1993: 298) provides a range of instances from Old English. The most convincing examples involve the verb magan, as is illustrated in example (998). This pattern is very similar to the German es kann sein-pattern, which is almost always interpreted in an epistemic way, as was illustrated by Doitchinov (2001: 119) and in examples (94)–(97) discussed in Section 2.2.1.5. Moreover, Doitchinov (2001) discusses individual occurrences of sculan and willan. Yet, their status appears to be less clear. (997)

Ther evangelio thar quit, theiz mohti wesan sexta zit¹ the Gospel there says this may-sbjv.pst be-inf sixth hour ‘The Gospel says at this point that it might have been the sixth hour.’

(998)

Swiθe eaθe θæt mæg beon θæt some men θencan. . . ² very easily that may be-inf that some men think ‘It may very well be that some men think. . . ’

As the data collected by Bolkestein (1980: 89–103, 123–133) show, the ambiguity of related verbs was established already in Latin: The necessity verbs debere ‘must’ and oportet ‘must’ could already alternatively be interpreted in a circumstantial and an epistemic manner. Accordingly, the development of ambiguous polyfunctional modal verbs in Germanic languages could be a result of language contact with Latin.

1 Otfrid II, 14, 9–10 (around 870), as quoted in Krause (1997: 95), translated by JM. 2 The Blickling Homilies 21.17 (around 980), as quoted in Denison (1993: 299).

7.1 Epistemic modal verbs in Early New High German | 543

Yet, regarding West Germanic languages, epistemic modal verbs remain rare until the early 16th century. As Müller (2001: 244) has observed, there are only a few convincing occurrences of epistemic modal verbs in the Prose Lancelot, which was written in the 13th or 14th century. He discussed one example of epistemic mögen, and one of reportative sollen. According to Müller (2001: 243), the scrutinised corpus contains 200,000 word form tokens. Likewise, Denison (1993: 298) has noticed that epistemic uses of modal verbs in English are only marginally developed in the Old and Middle English period. They are not systematically established until Early Modern English. In a similar vein, Fritz (1991: 45, 1997: 94) and Diewald (1999: 365) have demonstrated that epistemic modal verbs only became frequent in the course of the Early New High German period. According to these findings, the present study investigates Ulrich Schmid’s Neuwe Welt published in 1567, which is a travelogue of Portuguese discoverers travelling to India.³ The underlying corpus encompasses the preface plus the first chapter, which comprise in total 44,687 word form tokens. Regarding the genre, it should be seen more as a narrative text rather than a journalistic report. A journalistic report would rely on different sources or even just rumours, and in order to be objective it would evaluate them with regard to the preconceived convictions of the author. This is fairly different from a travelogue, which is normally founded on only one source of evidence: The sensory input and the immediate experience of the author himself. In contrast to the investigation carried out by Müller (2001: 244), there are at least seven unambiguously epistemic modal verbs among 44,000 word form tokens. The frequency of epistemic modal verbs in Neuwe Welt appears to be considerably higher than in the Prose Lancelot. As already in previous stages of West Germanic languages, the verb which is most frequently attested with an epistemic interpretation is moe gen (5 times), followed by koe nnen (1) and mue ssen (1). As in the case of the Old High German example (997) provided by Krause (1997: 95), an epistemic reading appears to become more likely if the relevant verb is specified for the subjunctive of the past. Only two occurrences are inflected for the indicative present (cf. 999–1000), one for the indicative past (cf. 1001), whereas four occurrences exhibit the subjunctive of the past (cf. 1002–1005). This shows that the subjunctive of the past facilitates an epistemic interpretation.

3 I am grateful to Christiane Wanzeck, who provided me with a hard copy of the original print from 1567.

544 | 7 On black magic: A diachronic explanation

(999)

Vnnd erstlich gibt er jhm sechtzig Kanons an gold / das ist ein and first gives he him sixty Kanons of gold that is a Mue ntz die also heist / vnd mag zusammen drey coin that alike be.called and may altogether three Portugalesischer Croisaden seyn / das were dreissig Portuguese Croisades be-inf that be-sbjv.pst thirty Ducaten.⁴ ducats ‘At first he gives him sixty golden Kanons, that so-called coin might value about thirty Portuguese Croisades or thirty ducats.’ (epistemic)

(1000)

Vnd welches er nach dieser sach gedae chte / das were and which he after this incident think-pst this be-sbjv.pst vrsach gnug / wie vil er jetzundt vielleicht auff vns reason enough how much he now maybe about us halten mag / das er vns hernach auch verachtet.⁵ think-inf may that he us then also disdains ‘And the thoughts which he had after this incident were reason enough that he would disdain us even afterwards – whatever he may think about us now.’ (epistemic)

(1001)

Vnd wie er im wenden war / da vername er sieben oder acht and as he at veer-inf was there noticed he seven or eight Blut Schiffe / die jm aus der Inseln mit auffgezogenen Segel blood ships the him from the islands with hoisted sails nachfuhren / vnnd mochten von des Nicolas de Cocillo schiff followed and may-pst from the-gen Nicolas de Cocillo ship ein grosse meil wegs weit seyn.⁶ a large mile way-gen away be-inf ‘And as he was about to veer, he noticed seven or eight blood ships that followed him from the island with hoisted sails and they may have been about one large mile away from the Nicolas de Cocillo’s ship.’ (epistemic)

(1002)

Des Sontags / vngefehrlich vmb Vesper zeit / kamen die the-gen sunday-gen about at vespers time came the vnsern bey drey Inseln / die alle sehr klein waren / vnd ours by three islands the all very small were and

4 Ulrich Schmid, Neuwe Welt, p. 19b, (1567). 5 Ulrich Schmid, Neuwe Welt, p. 21b, (1567). 6 Ulrich Schmid, Neuwe Welt, p. 7b, (1567).

7.1 Epistemic modal verbs in Early New High German |

545

moe chte eine von den andern vielleicht vier meil weit may-sbjv.pst one from the others maybe four miles away seyn.⁷ be-inf ‘They arrived on sunday at vespers time at three islands that were all very small and that might have been away four miles from each other.’ (epistemic)

(1003)

Es moe chte wol seyn / daß sie etwan ein wenig auß it may-sbjv.pst well be-inf that she eventually a little of dem weg gefahren weren / vnd darumb kae men sie the way travel-ppp be-sbjv.pst and therefore come-sbjv.pst they in das wetter / das regiert gewoehnlich vmb die zeit⁸ in the weather rel.prn reigns usually around the time ‘It may be that they eventually deviated a little bit from the course and therefore they were exposed to weather which is usual at that period.’ (epistemic)

(1004)

[...] hielte er dafue r / es mue ste der Koe nig auß Portugal thought he that it must-sbjv.pst the king of Portugal ein dapfferer geherzter Mann seyn.⁹ a brave hearted man be-inf ‘He thought that the King of Portugal must be a bold and brave man.’

(1005)

(epistemic)

darab der Oberst sehr froh ward / denn er gedachte bey about.that the colonel very happy got for he thought by jhm selbst / dieweil er zu Leuten kommen were / die him self because he to people come be-sbjv.pst the etlicher massen Schiffung hetten / so koe ndte Indien nicht many size navigation had thus can-sbjv.pst India neg mehr weit seyn¹⁰ more far be-inf ‘Therefore the colonel became very happy, for he thought that since he came to people with the knowledge of navigation India could not be that far any more.’ (epistemic)

As illustrated in example (1000), the concessive epistemic interpretation of mag was already available in the 16th century. In the example provided above, it occurs in a clause which has to be analysed as a concessive conditional along the lines suggested by König and Auwera (1988: 118), as it is introduced by a free choice

7 Ulrich Schmid, Neuwe Welt, p. 7b, (1567). 8 Ulrich Schmid, Neuwe Welt, p. 36a, (1567). 9 Ulrich Schmid, Neuwe Welt, p. 15a, (1567). 10 Ulrich Schmid, Neuwe Welt, p. 7a, (1567).

546 | 7 On black magic: A diachronic explanation

Tab. 7.1: Circumstantial modal verbs with stative complements – in Schmid’s Neuwe Welt (1567) circumstantial modal verb moe gen (91) koe nnen (154) mue ssen (33) wollen (307) sollen (179) due rfen (22)

stative complements seyn (3), Kimean (3; wissen, leiden) seyn (3), Kimean (4; haben, wissen) seyn (3), Kimean (2; haben) seyn (7), Kimean (7; haben) seyn (9), Kimean (3; freuen, halten, haben) seyn (1)

percentage 6.59 % 4.55 % 15.15 % 4.56 % 6.70 % 4.55 %

item wie vil. The epistemic interpretation of mag in the pattern above is even more plausible, considering that Menedéz-Benito (2010: 33) has shown that free choice items have an affinity to occur with possibility modal operators which can also be epistemic. Moreover, it deserves closer attention that six out of the seven epistemic modal verbs attested in this corpus occur in non-embedded main clauses. There is only one epistemic modal verb which occurs in an embedded clause (cf. 1004). Yet, the respective complement clause lacks a subordinative conjunction and exhibits V2 order, which is generally regarded as a characteristic of non-integrated complement clauses, as was pointed out by Antomo and Steinbach (2010: 12), among others. Rather, it should be analysed as a clause which is adjoined to the matrix VP. Apart from that, the epistemic modal verbs mentioned above share another essential characteristic. All of them select stative predicates as complements: most notably, the copula seyn ‘be’, as illustrated in examples (999) and (1001)–(1005), or the mental state predicate halten ‘consider’, as shown in example (1000). Many of them refer to a state which cannot be changed, or is not likely to change. In contrast, Maché (2008: 401) has shown that circumstantial modal verbs in Schmid’s Neuwe Welt occur by far less frequently with stative complements. The precise figures are illustrated in Table 7.1. This is in accordance with the observations made by Abraham (1991), Abraham (2001) and Abraham (2005) and Leiss (2002), who have demonstrated that circumstantial modal verbs have a preference for predicates with an eventive semantics. Aside from the unambiguous occurrences of epistemic modal verbs, the corpus contains roughly ten more instances of modal verbs which could be interpreted in an epistemic manner under certain conditions. Interestingly, they are characterised by a couple of preferences. As Maché (2008: 393) has already pointed out, all of these ambiguous occurrences select eventive predicates as a complement. Furthermore, they mostly occur embedded under attitude predicates. Finally, these modal verbs typically bear past subjunctive morphology. Maché

7.1 Epistemic modal verbs in Early New High German | 547

(2008: 390) has already suggested that the choice of the subjunctive of the past morphology could be triggered by the syntactic context. As he has illustrated, moe gen occurs in embedded clauses in 76 out of 99 cases, in which it is mostly specified for the subjunctive of the past. In what follows, some of the ambiguous instances will be inspected more closely. The most likely interpretation of moe gen in example (1006) is an ability reading. This corresponds to the manner in which it was used in Old High German. At this period, the verb was frequently employed to attribute an ability to the subject referent. Moreover, the modal verb is realised as an IPP participle which is embedded by a perfect tense auxiliary with subjunctive of the past morphology. This is an environment which is not likely to host epistemic operators. (1006)

Bontaibo verwunderte sich sehr / wie sie zu wasser hetten Bontaibo wondered refl very how they at water had kommen moe gen / un¯ fragten jn was sie sucheten / come-inf may-ppp(ipp) and asked him what they searched weil sie so weit gefahren weren:¹¹ because they that far travel were ‘Bontaibo was very surprised at what had enabled them to come across the sea and asked them what they were looking for, since they were travelling so far:’

Furthermore, there are usages of moe gen which are fairly likely to be interpreted in a circumstantial manner. Some of them are reminiscent of possibility modal verbs which quantify over situations, such as the English verb can, which is discussed in Section 3.2. In examples (1007)–(1010), the speaker expresses that he is aware of the possibility that the state of affairs expressed by the proposition can happen, under certain conditions. (1007)

Denn der Haf¯e wer besser daselbst / denn zu Calecut / for the harbour be-sbjv.pst better there than at Calecut da die seit sehr gefehrlich ist / vnnd die Schiff moe chten where the coast very dangerous is and the ship may-sbjv.pst vielleicht daselbst verderben.¹² maybe there perish-inf ‘For that harbour was better than the one at Calecut where the coast is so perilous that the ship could maybe get lost.’

11 Ulrich Schmid, Neuwe Welt, p. 20a, (1567). 12 Ulrich Schmid, Neuwe Welt, p. 21a, (1567).

548 | 7 On black magic: A diachronic explanation

(1008)

denn er wuste nicht / was jme etwan begegnen for he knew neg rel.prn him eventually happen-inf moe chte.¹³ may-sbjv.pst ‘For he didn’t know what could (perhaps) happen to him.’

(1009)

Auch solte er bedencken / wie es nach sein¯e todt jnen also shall-sbjv.pst he reflect how it after his death them allen ergehen moe chte.¹⁴ all go-inf may-sbj.pst ‘Moreover he should imagine how their fate would be when he died.’

(1010)

denn sie furchteten / es moe chte jhnen begegnen for they feared it may-sbjv.pst them happen-inf was jhn S. Helenen Hafen begegnet war¹⁵ rel.prn them S. Helena port happen-ppp was ‘For they were afraid that the same could (possibly/presumably) happen to them that happened to them at port S. Helena.’

Summing up, most of the examples (1006)–(1010) are fairly likely to be interpreted in a circumstantial manner. As a consequence, those instances which unambiguously involve epistemic modal verbs behave in a fairly uniform manner: They all select some sort of stative predicate which, in many cases, refers to a state which is not likely to be changed. Regarding the situation of modal verbs in Early New High German, it appears to be plausible that the first epistemic modal verbs in history originally selected stative predicates. The circumstantial modal verbs found in Ulrich Schmid’s Neuwe Welt dominantly occur with eventive predicates as complements, which conforms to the hypothesis formulated above, according to which circumstantial modal verbs are event modifiers, as illustrated in Section 3.2. In opposition to that, epistemic modal verbs are frequently attested with stative predicates which refer to a state which is not likely to change. Accordingly, the grammaticalisation of modal verbs could be seen as a change from event modifiers into propositional (or speech act) modifiers. The following section will provide a detailed scenario of how epistemic modifiers have most likely evolved.

13 Ulrich Schmid, Neuwe Welt, p. 21a, (1567). 14 Ulrich Schmid, Neuwe Welt, p. 21a, (1567). 15 Ulrich Schmid, Neuwe Welt, p. 5b, (1567).

7.2 The Rule of Accommodation as a driving force of language change | 549

7.2 The Rule of Accommodation as a driving force of language change At this point, it is important to notice that circumstantial modal verbs can occasionally be combined with typical stative predicates such as sein or haben. Likewise, epistemic modal verbs are not restricted to stative predicates but can sometimes also select eventive predicates. At first glance, it appears that these configurations are in conflict with the hypothesis outlined in Section 3.2, according to which circumstantial modal operators are event modifiers – why should an event modifier modify a state? There are several authors, such as Kratzer (1995: 148) and Maienborn (2003: 178, 193, 216), who have suggested that highly abstract stative predicates can be transformed into more eventive predicates by means of pragmatic mechanisms. As they assume, there is a class of stative predicates (Individual Level Predicates in terms of Carlson (1977), Kratzer (1995) and Diesing (1992) or Kimean State Predicates in terms of Maienborn (2003)), which do not involve a Davidsonian event argument. Following their ideas, an event modifier should not modify such a stative predicate because it lacks an event argument. Correspondingly, a circumstantial modal verb is restricted to the modification of predicates which involve an event argument. If it selects a stative predicate such as sein, this would result in a configuration which violates the selectional restrictions imposed by the circumstantial modal. As suggested by Lewis (1979: 172) and Kratzer (1981: 61), there is a way of providing the required type of complement if it is missing. This rule can be applied under certain circumstances and is known as the Rule of Accommodation: Rule of Accommodation If the utterance of an expression requires a complement of a certain kind to be correct, and the context just before the utterance does not provide it, then ceteris paribus and within certain limits, a complement of the required kind comes into existence.

As Kratzer (1981: 61) remarks, this rule “is black magic, but it works in many cases.” It does in the case of circumstantial modal operators. Maienborn (2003: 178, 193, 216) has proposed two pragmatic mechanisms that can supply a stative predicate with an event argument if one is missing. In her reasoning, any Kimean state predicate lacks an event argument in the lexicon, as illustrated in Maienborn (2003: 106). Accordingly, these predicates by default refer to temporally unbound states. They can be delimited if a specific event argument is supplied. This can be provided by the Temporariness Effect, which introduces temporal boundaries for the state, as exemplified in She was tired. Moreover, the Agentivity Effect can

550 | 7 On black magic: A diachronic explanation

turn a state into a volitionally controlled action, which is, in turn, temporally restricted: Dafna is being polite. In other words, any predicate which is selected by a circumstantial modal verb will be interpreted as a predicate which refers to a temporally restricted state or event. As was shown in Sections 2.2.1.5 and 3.3, there are contexts in which the communicative effect of an epistemic possibility verb is almost identical to the communicative effect of a circumstantial possibility verb. Correspondingly, it is a challenging endeavour for the addressee to guess which alternative the speaker has realised and intended. This type of context is the typical situation in which reanalysis takes place. There are two possible alternatives: The operator under consideration could either be interpreted as a practical possibility verb supported by pragmatic repair mechanism, or as an epistemic modal verb. As the latter interpretation does not require a pragmatic repair mechanism in order to be acceptable, it is less complex. As was demonstrated by Lightfoot (1979: 375), Roberts and Roussou (1999: 1022) and Roberts (2003: 16) language learners tend to assume the least complex linguistic analysis for the input to which they are exposed. Accordingly, any language learner would prefer the analysis built on the epistemic modal verb over the analysis which involves a circumstantial modal verb that is combined with an illicit complement, requiring an opaque repair mechanism. At this point, the question arises what role the subjunctive of the past plays in this scenario. As was seen, the majority of epistemic modal verbs in earlier stages of German seem to be specified for the past. The use of the subjunctive might be an indicator of decreased commitment. In Sections 2.2.1.5 and 2.2.2.5, it was demonstrated that the subjunctive of the past on epistemic modal verbs indicates that some of the premises on which the evaluation is based are not verified.

7.3 Summary Of course, the analysis outlined in the present chapter remains very sketchy. In the near future, when large scale electronic corpora for historical data will be available, it will be possible to provide an analysis which is based on more solid empirical foundations. This section indicates what direction it is fruitful to pursue for future research on grammaticalisation. The type of the embedded predicate appears to play a key role in the development of epistemic modal verbs. As shown throughout the study, circumstantial modal operators are most likely to be seen as event modifiers, while epistemic modal operators are operators which act on the clause level. Accordingly, the grammaticalisation of epistemic modality can be captured as a change from event modification to clausal modification. This confirms the

7.3 Summary

|

551

observation made by Abraham (1991), Abraham (2001) and Abraham (2005) and Leiss (2002), who argue that aspectual semantics of the embedded predicate play a crucial role in the grammaticalisation of epistemic modality. Moreover, it was demonstrated in Chapter 6 that epistemic modal verbs can still be considered elements of category ‘verb’. Accordingly, the process of grammaticalisation under discussion can be considered as a development which does not affect the syntactic category of the underlying verb. The essential clue for understanding this development is that circumstantial modal operators are capable of selecting stative predicates as long as they can be reinterpreted as events by means of a pragmatic repair mechanism. As long as language learners are able to detect this mechanism, the modal operators will be attributed a circumstantial interpretation. In some contexts, however, the communicative effect of a circumstantial modal operator and an epistemic one is almost identical. It can thus happen that the application of the repair mechanism becomes too opaque for the language learner. As a consequence, any learner would prefer an epistemic interpretation, which is the less complex alternative in this case.

8 Summary This must be the end. Finally, all of the potential modal verbs in German have been discussed in some detail, considering the most important findings from several centuries of research and grounded on broad empirical data. The result is the first comprehensive corpus based description of the so-called modal verbs in German. The underlying corpus is the archive W of the DeReKo corpus, which was compiled at the Institut für Deutsche Sprache in Mannheim, and which encompassed about 2 billion word form tokens at the period when this research was carried out. In what follows, a short overview of the most important findings will be given. In Chapter 2, it was suggested that there are two possibilities to define the individual class of modal verbs: a strong definition, which includes all possible uses of each individual modal verb lexeme, and a weak one, which only considers their epistemic interpretations. In the course of the study, it was shown that only the weak definition is plausible. The epistemic interpretations of each traditional modal verb behave in a fairly homogeneous way. As a consequence, they can be grouped into a class. By contrast, their circumstantial interpretations are very idiosyncratic. While it is possible to consider the circumstantial readings of single lexical items as a class, this approach would fail for the entire set of circumstantial modal verbs. As it is impossible to subsume the circumstantial modal verbs under a homogeneous class, the strong definition of modal verbs is not applicable to the traditional six elements in German. Using a weak definition of modal verbs, two further elements have to be integrated: brauchen and werden. Moreover, it has been shown that epistemic dürfte has to be as regarded an independent lexical item. It was moreover shown in Section 2.3 that the contradictions originate in the terminology used. In the grammars of the 17th century, verbs with a preterite present inflection pattern were subsumed under a particular class. At that time, the motivation was a purely morphological one. Only in the 19th century did grammarians additionally try to provide a functional motivation for this class. Unfortunately, there was always a mismatch between those verbs with exceptional morphology, and those verbs with a remarkable function. Accordingly, early definitions which were only based on a single motivation result in a much more homogeneous class than ‘mixed’ definitions do. Chapter 3 was dedicated to the examination of the nature of epistemic modality. Two characteristics were identified: firstly, epistemic modifiers are capable of modifying predications consisting of an identified individual, and a predicate which refers to a state that is not likely to change, or a predicate that refers to an

https://doi.org/10.1515/9783110411027-008

8 Summary

|

553

event in the past. Secondly, the use of an epistemic modifier indicates that the embedded proposition is not part of the deictic centre’s knowledge. Chapter 4 dealt with the environments from which epistemic modal verbs are banned. Among the 21 non-canonical environments which have been suggested in the literature, there are only eight in which epistemic modal verbs could not be found: (i) they do not occur with verbless directional phrase complements, (ii) they cannot be separated from their infinitive complements in wh-clefts, (iii) they do not undergo nominalisation, (iv) they are banned from adverbial infinitives, and they cannot be embedded under (v) circumstantial modal verbs, (vi) predicates of desire, (vii) imperative operators, or (viii) optative operators. Furthermore, it was demonstrated in Section 4.22 that the distinction between subjective and objective epistemic modality is misleading, and that it cannot account for the distribution of epistemic modal verbs. In particular, the alleged characteristics of objective modality only apply to individual verbs, but never to all verbs which are considered as ‘objective’ epistemic. Most of these patterns can be treated more efficiently as (‘subjective’) epistemic instances, which exhibit verbspecific idiosyncrasies. The remaining cases are circumstantial modal verbs. Moreover, there is compelling evidence that epistemic modal verbs should be analysed as a part of the proposition, as they contribute to the truth conditions, as pointed out in Section 4.23. In Chapter 5, it was shown that reportative modal verbs are systematically more flexible with respect to the non-canonical environments discussed above. In contrast to their epistemic counterparts, reportative modal verbs are attested: in nominalisations (iii), in adverbial infinitives (iv), and embedded under tense auxiliaries (v) as well as optative operators (viii). As demonstrated in Chapter 6, the behaviour of epistemic and reportative modal verbs can most efficiently be captured in terms of anchoring conditions regarding the deictic centre. These operators introduce variables for the deictic centre which have to be bound by an appropriate attitude holder. Generally, this variable will be anchored to the most local argument which can be interpreted as an attitude holder. In case there is none, it will be tied to the most salient referent of the speech act – otherwise the variable will be left uninstantiated. There are other operators, such as circumstantial modal operators or nominalisation operators, which fail to embed linguistic structures that contain a free variable for a deictic centre. As a consequence, epistemic modal verbs can be described as elements of the category ‘verb’ which introduce a variable for a deictic centre that needs to be identified. Thus, they need not be analysed as affix-like elements merged in a functional projection. Subsequently, Chapter 7 provided a scenario which can explain how epistemic modal operators emerge. Being event modifiers, circumstantial modal

554 | 8 Summary

operators are restricted to the modification of events. Yet, they occasionally occur with stative predicates – as long as they can be reinterpreted as events by applying a pragmatic repair mechanism. If the application of this mechanism becomes too opaque, language learners are likely to reanalyse these modifiers as epistemic modal operators. This happens, in particular, in contexts where circumstantial and epistemic modal operators yield a similar communicative effect. As was seen in Chapter 6, epistemic modal verbs are elements of category V. Accordingly, the grammaticalisation of epistemic modal verbs can be considered as a process which leaves the syntactic category of the verb intact. Considering the findings summarized above, we are finally in a position to answer the questions raised in Section 1.1: What is the nature of modal verbs? As for German, the strong definition of modal verbs results in a very imprecise concept, which has no value for research. By contrast, the weak definition yields a homogeneous class which encompasses all epistemic modal verbs. Accordingly, it is recommended to either use the term modal verb with great care, or else abandon it. What is the nature of epistemic modifiers? Epistemic modifiers are clausal modifiers which indicate that the embedded proposition is not part of the deictic centre’s knowledge. Hence, they contribute to the truth conditions, as a deictic centre is required whose knowledge does not contain the embedded proposition. What triggered grammaticalisation? In virtue of their nature as event modifiers, circumstantial modal verbs are restricted to the modification of events. Sometimes, however, they are combined with stative predicates which are interpreted as events by means of a pragmatic repair mechanism. As long as language learners are able to detect this repair mechanism, they will analyse the relevant patterns as circumstantial modification. As soon as the application of the repair mechanism becomes too opaque, the language learner will reinterpret them as instances of epistemic modification, which appears to be the less complex, and more plausible, configuration in this situation.

Primary sources Alexis, Willibald (1894). Der Werwolf. Berlin: Otto Janke. Ausführliche Arbeit von der Teutschen HaubtSprache (1663). Braunschweig: Christoff Friedrich Zilligern. Berg, Franz (1793). Predigten über die Pflichten der höheren und aufgeklärten Stände bey den bürgerlichen Unruhen unserer Zeit. Wirzburg: Stahel. Bergius, Peter Jonas (1794). Von Obstgärten und deren Beförderung in Schweden. Leipzig: Gräffische Buchhandlung. Chomsky, Noam (2002). Offene Wunde Nahost. Translated by Michael Haupt. Hamburg: Europa Verlag. Cochläus (Vogelgesang), Johannes (1900). Ein heimlich Gespraech von der Tragedia Johannis Hussen. Edited by Hugo Holstein. Halle a. Saale: Max Niemeyer. Cramer, Johann Andreas (1758). Der Nordische Aufseher. Kopenhagen: Johann Benjamin Ackermann. Dostojewskij, Fjodor Michailowitsch (2003). Aufzeichungen aus dem Kellerloch. Translated by Svetlana Geier. Frankfurt: Fischer. Dostojewskij, Fjodor Michailowitsch (2009). Der Idiot. Translated by Svetlana Geier. Frankfurt: Fischer. Eschenbach., Wolfram von (1998). pârzival edited by Karl Lachmann. 6th ed. De Gruyter. Fichte, Johann Gottlieb (1798). Das System der Sittenlehre. Vol. II. Leipzig: Meiner. Gellert, Christian Fürchtegott (1747a). ‘Das Orakel’. In: Lustspiele. Leipzig: Johann Wendler. Gellert, Christian Fürchtegott (1747b). ‘Die Betschwester’. In: Lustspiele. Leipzig: Johann Wendler. Gellert, Christian Fürchtegott (1747c). ‘Die Zärtlichen Schwestern’. In: Lustspiele. Leipzig: Johann Wendler. Goethe, Johann Wolfgang von (1991). Faust - der Tragödie erster Teil. Edited by Erich Trunz. Beck. Gotter, Friedrich Wilhelm (1772). ‘Elektra’. In: Gedichte. Vol. II. 45. Grimm, Jacob and Wilhelm Grimm (1837). ‘Die zertanzten Schuhe’. In: Kinder und Hausmärchen. Große Ausgabe. Günther, Johann Christian (1720). ‘Ode an Herrn Marckard von Riedenhausen Ivris Vtrivsqve Cvltor’. In: Johann Christian Günther Werke. Ed. by Reiner Bölhoff. Frankfurt: Deutscher Klassiker Verlag. Gustav, Hugo (1829). Beyträge zur civilistischen Bücherkenntnis der letzten vierzig Jahre. Berlin. August Mylius. Hauptmann, Gerhart (1906). ‘Die Weber’. In: Gesammelte Werke. Berlin: Fischer. Heine, Heinrich (1828). ‘Reise nach Italien, Abschnitt IV’. In: Morgenblatt für gebildete Stände 20. Hemmer, Johann Jakob (1769). Abhandlung über die deutsche Sprache zum Nutzen der Pfalz. Mannheim: Akademische Schriften. Heyse, Paul (1862). ‘Auf der Alm’. In: Neue Novellen. Vol. 4. Berlin: Wilhelm Hertz. Jörg, Edmund and Franz Binder (1870). Historisch-politische Blätter für das katholische Deutschland. Vol. 26. München: Literarisch-Artistische Anstalt.

https://doi.org/10.1515/9783110411027-009

556 | Primary Sources

Kafka, Franz. (1994). ‘Die Verwandlung’. In: Franz Kafka. Gesammelte Werke. Drucke zu Lebzeiten. Ed. by Wolf Kittler, Hans-Gerd Koch and Gerhard Neumann. Frankfurt/Main: S. Fischer. Kleist, Heinrich von (1995). ‘Der Zebrochene Krug’. In: Sämtliche Werke. Ed. by Roland Reuß and Peter Staengele. Stroemfeld/Roter Stern. Kramer, Matthias. (1702b). Das herrlich-Grosse Teutsch-Italiänische Dictionarium. Nürnberg: Johann Andrea Endters Söhne. Kraus, Karl (1925). Die Fackel. 2nd ed. München: Kösel Verlag. Kraus, Walther (1983). ‘Das Gericht über Orest bei Aischylos’. In: Festschrift Robert Muth. Ed. by . Paul Händel. Innsbrucker Beiträge zur Kulturwissenschaft. Innsbruck: Institut für Sprachwissenschaft der Universität Innsbruck. Lenz, Jakob Michael Reinhold (1774). Der Hofmeister. Leipzig: Weygandsche Buchhandlung. Lessing, Gotthold Ephraim (1766). Laokoon: oder über die Grenzen der Mahlerey und Poesie. Christian Friedrich Voss. Lessing, Gotthold Ephraim (1767a). Hamburgische Dramaturgie. J.H. Cramer. Lessing, Gotthold Ephraim (1767b). ‘Minna von Barnhelm’. In: Lustspiele. Zweyter Theil. Ed. by Gotthold Ephraim Lessing. Christian Friedrich Voß. Lessing, Gotthold Ephraim (1768). Hamburgische Dramaturgie. J.H. Cramer. Lohenstein, Daniel Casper von (2005a). Agrippina. Vol. 2. Sämtliche Werke. Historisch kritische Ausgabe edited by Lothar Mundt. Berlin: De Gruyter. Lohenstein, Daniel Casper von (2005b). Epicharis. Vol. 2. Sämtliche Werke. Historisch kritische Ausgabe edited by Lothar Mundt. Berlin: De Gruyter. Lohenstein, Daniel Casper von (2008). Ibrahim Bassa. Vol. 1. Sämtliche Werke. Historisch kritische Ausgabe edited by Lothar Mundt. Berlin: De Gruyter. Moscherosch, Johann Michael. (1640). Alamodischer Politicus. Cölln: Andreas Bingen. Moscherosch, Johann Michael. (1647). Alamodischer Politicus. Cölln: Andreas Bingen. Münchhausen, Otto Freiherr von (1771). Monathliche Beschaeftigungen fuer einen Baum= und Plantagen=Gaertner. Hannover. Palm, Kurt (2005). Der Mozart ist fett und wohlauf. Wien: Löcker. Rückert, Friedrich (1821). ‘Nachklang’. In: Morgenblatt für gebildete Stände, 243. Schaidenreisser, Simon (1537). Odyssea. Translation of Homer’s epos. Augsburg: Alexander Weissenhorn. Schiller, Friedrich von (1823). Geschichte des dreissigjährigen kriegs, zweyter teil. Vol. 54. Werke II. Carlsruhe: Bureau deutscher Klassiker. Schmid, Ulrich (1567). Neuwe Welt. Franckfurt am Mayn: Siegmund Feyrabend. Schulze, Ernst (1818). Cäcilie. Leipzig: Brockhaus. Schwaben, Johann Joachim (1751). Herrn Johann Christoph Gottscheds Gedichte. Leipzig: Breitkopf. Tatian edited by Edmund Sievers (1961). 2nd ed. Darmstadt: Wissenschaftliche Buchgesellschaft. Ullmann, Carl (1829). Theologische Studien und Kritiken. Vol. II. 1. Hamburg: Friedrich Perthes. Verwandlungen. Translation of Ovid’s Metamorphoses (1798). Berlin: Friedrich Vieweg der Ältere. Vogelweide, Walter von der (1996). Leich, Lieder, Sangsprüche. Edited by Karl Lachmann. 14th ed. Berlin: De Gruyter. Voß, Johann Heinrich (1781). Odyssee. Translation of Homer’s epos. Hamburg. Voß, Johann Heinrich (1793). Illias. Translation of Homer’s epos. Hamburg.

Primary Sources |

557

Weinhold, Moritz, ed. (1862). Achtundvierzig Briefe von Johann Gottlieb Fichte und seinen Verwandten. Leipzig: Grunow. Winkelmanns, Johann (1764). Geschichte der Kunst des Alterthums. Dresden: bei Walther. Zwengel, Johann Peter (1568). New Groß Formular und vollkommlich Cantzlei Buch von den besten und außerlesenen Formularien aller deren Schrifften/ so in Chur und Fürstlichen / auch der Grauen / Herren unnd anderen fürnemen Cantzleyen / Auch sonst in den Ampten unnd Ampts händeln / Deßgleichen under dem gemeinenMan / allerley fürfallendert geschäfft halben / bräuchlich seindt – Reuerßbrieff vber bestandnen Bawhofe/sampt deren Güter. Frankfurt: Egenolffs Erben.

References Abraham, Werner (1991). ‘Modalverben in der Germania’. In: Begegnung mit dem Fremden: Grenzen – Traditionen – Vergleiche. Akten des VIII. Internationalen Germanisten Kongress in Tokio. Ed. by Eijiro Iwasaki. Vol. 4. München: iudicium, pp. 109–118. Abraham, Werner (2001). ‘Modals: towards explaining the ‘epistemic non-finiteness gap’.’ In: Modalität und Modalverben im Deutschen. Ed. by Reimar Müller and Marga Reis. Hamburg: Buske, pp. 7–36. Abraham, Werner (2002). ‘Modal verbs: Epistemics in German and English’. In: Modality and its interaction with the verbal system. Ed. by Sjef Barbiers, Frits Beukema and Wim van der Wurff. Vol. 47. Linguistik Aktuell. Amsterdam: Benjamins, pp. 19–50. Abraham, Werner (2003). ‘Canonic and non-canonic deliberations about epistemic modality: its emergence out of where? On an extended notion of grammatic(al)ization’. In: Germania et alia. A Linguistic Webschrift for Hans den Besten. Ed. by Jan Koster and Henk van Riemsdijk. University of Amsterdam. Abraham, Werner (2004). ‘The grammaticalization of the infinitival preposition - towards a theory of ’grammaticalizing reanalysis’’. In: Journal of Comparative Germanic Linguistics 7.2, pp. 111–170. Abraham, Werner (2005). ‘Event Arguments and modal verbs’. In: Event Arguments: Foundations and Applications. Ed. by Claudia Maienborn and Angelika Wöllstein. Vol. 501. Reihe Linguistische Arbeiten. Niemeyer, pp. 243–276. Abraham, Werner (2011). ‘Traces of Buhler’s semiotic legacy in modern linguistiscs’. In: Theory of language. Ed. by Karl Bühler. 2nd ed. Amsterdam: John Benjamins. Chap. Preface, pp. xiii–xlvii. Adelung, Johann Christoph (1782). Umständliches Lehrgebäude der deutschen Sprache: zur Erläuterung der deutschen Sprache für Schulen. Leipzig: Breitkopf. Adelung, Johann Christoph (1793). Grammatisch-kritisches Wörterbuch der Hochdeutschen Mundart: A-E. 2nd ed. Vol. 1. Leipzig: Breitkopf und Härtel. Adelung, Johann Christoph (1798). Grammatisch-kritisches Wörterbuch der Hochdeutschen Mundart: M-Scr. 2nd ed. Vol. 3. Leipzig: Breitkopf und Härtel. Adelung, Johann Christoph (1801). Grammatisch-kritisches Wörterbuch der Hochdeutschen Mundart: Seb-Z. 2nd ed. Vol. 4. Leipzig: Breitkopf und Härtel. Adéwo.lé, Fé.mi (1990). ‘Gbó.dò. ‘must’: analysis of a Yorùbá modal verb’. In: Journal of West African Languages 20.1, pp. 73–82. Adéwo.lé, Fé.mi O. (1993). ‘A semantic analysis of a Yorùbá modal verb’. In: Ifè.: Annals of the Institute of Cultural Studies 4, pp. 37–45. Aijmer, Karin (1978). ‘Epistemic modal auxiliaries in English’. In: Papers from the fourth Scandinavian Conference of Linguistics. Ed. by Kirstin Gregersen, Hans Basbøll and Jacob Mey. Odense, pp. 161–168. Alban, Doctor (1992). ‘It’s my lifè.’. PhD thesis. Stockholm: Eurodance. Aldenhoff, Jules (1962). ‘Der Ersatzinfinitiv im heutigen Deutschen’. In: Revue des langues vivantes 28, pp. 195–217. Allard, Francis X. (1975). A structural and semantic analysis of the German modal ‘mögen’. Vol. 6. Stanford German studies. Frankfurt/Main: Peter Lang. Anand, Pranav and Valentine Hacquard (2013). ‘Epistemics and attitudes’. In: Semantics & Pragmatics 6.8, pp. 1–59. https://doi.org/10.1515/9783110411027-010

560 | References

Antomo, Mailin and Markus Steinbach (2010). ‘Desintegration und Interpretation: Weil-V2Sätze an der Schnittstelle zwischen Syntax, Semantik und Pragmatik’. In: Zeitschrift für Sprachwissenschaft 29.1, pp. 1–37. Asbach-Schnitker, Brigitte (1977). ‘Die Satzpartikel ‘wohl’’. In: Aspekte der Modalpartikel. Ed. by Harald Weydt. Tübingen: Niemeyer, pp. 38–61. Askedal, John Ole (1989). ‘Über den Infinitiv ohne bzw. mit ‘zu’ im heutigen Deutsch: Klassenbildung regierender Lexeme und Hauptzüge der Distribution.’ In: Deutsch als Fremdsprache 26.1,2, pp. 2–7, 103–105. Askedal, John Ole (1991). ‘Ersatzinfinitiv/Partizipersatz und Verwandtes. Zum Aufbau des verbalen Schlussfeldes in der modernen deutschen Standardsprache’. In: 19, pp. 1–23. Askedal, John Ole (1997a). ‘‘Brauchen’ mit Infinitiv. Aspekte der Auxiliarisierung’. In: Jahrbuch der ungarischen Germanistik, pp. 53–68. Askedal, John Ole (1997b). ‘‘drohen’ und ‘versprechen’ als sogenannte Modalitätsverben in der deutschen Gegenwartssprache’. In: Deutsch als Fremdsprache 34, pp. 12–19. Askedal, John Ole (1998). ‘Satzmustervariation und Hilfsverbproblematik beim deutschen Verb ’scheinen’’. In: Deutsche Grammatik - Thema in Variationen. Festschrift für Hans-Werner Eroms zum 60. Geburtstag. Ed. by Karin Donhauser and Ludwig Eichinger. Heidelberg: Winter, pp. 49–74. Auwera, Johan van der, Andreas Ammann and Saskia Klindt (2005). ‘Modal polyfunctionality and Standard Average European’. In: Modality. Studies in Form and Function. Ed. by Alex Klinge and Henrik Høeg Müller. London: Equinox, pp. 247–272. Axel, Katrin (2001). ‘Althochdeutsche Modalverben als Anhebungsverben’. In: Modalität und Modalverben im Deutschen. Ed. by Marga Reis and Reimar Müller. Hamburg: Buske, pp. 37–60. Barbiers, Sjef (1995). ‘The syntax of interpretation’. PhD thesis. Univeristeit Leiden. Barbiers, Sjef (2002). ‘Modality and polarity’. In: Modality and its interaction with the verbal system. Ed. by Sjef Barbiers, Frits Beukema and Wim van der Wurff. Vol. 47. Linguistik Aktuell. Benjamins, pp. 51–73. Bartsch, Renate (1972). Adverbialsemantik. Frankfurt/Main: Athenäum. Basler, Otto (1935). Der große Duden: Grammatik der deutschen Sprache; eine Anleitung zum Verständnis des Aufbaus unserer Muttersprache. Zürich: Fretz & Wasmuth. Bauer, Friedrich (1850). Grundzüge der Neuhochdeutschen Grammatik für die unteren und mittleren Klassen höherer Bildungsanstalten. Nördlingen: C.H. Beck. Bauer, Friedrich (1870). Grundzüge der Neuhochdeutschen Grammatik für die unteren und mittleren Klassen höherer Bildungsanstalten. 12th ed. Nördlingen: C.H. Beck. Bauer, Friedrich and Konrad Duden (1887). Grundzüge der Neuhochdeutschen Grammatik für höherer Bildungsanstalten und zur Selbstbelehrung für Gebildete. 20th ed. Nördlingen: C.H. Beck. Bauer, Heinrich (1827). Vollständige Grammatik der neuhochdeutschen Sprache. Vol. 1. Berlin: G. Reimer. Bauer, Heinrich (1832). Vollständige Grammatik der neuhochdeutschen Sprache. Vol. 4. Berlin: G. Reimer. Bech, Gunnar (1949). ‘Das Semantische System der Modalverben’. In: Travaux du Cercle Linguistique de Copenhague VI, pp. 3–46. Bech, Gunnar (1951). Grundzüge der semantischen Entwicklungsgeschichte der hochdeutschen Modalverba. Vol. 6. Historisk-filologiske meddelelser 32. Copenhagen: Munksgaard.

References |

561

Bech, Gunnar (1955). Studien über das deutsche Verbum infinitum. Vol. 35-36. Historiskfilologiske Meddelelser 2/6. Kopenhagen: Munksgaard. Bech, Gunnar (1963). ‘Grammatische Gesetze im Widerspruch’. In: Lingua 12, pp. 291–299. Becker, Karl Ferdinand (1836). Ausführliche deutsche Grammatik als Kommentar der Schulgrammatik. 2nd ed. Vol. 1. Frankfurt am Main: G. F. Kettembeil. Becker, Karl Ferdinand (1841). Organism der Sprache. 2nd ed. Frankfurt/Main: Kettembeil. Becker, Karl Ferdinand (1842). Ausführliche deutsche Grammatik als Kommentar der Schulgrammatik. revised 2. Vol. 1. Frankfurt am Main: G. F. Kettembeil. Behaghel, Otto (1924). Deutsche Syntax II: die Wortklassen und Wortformen B/C. Adverbium. Verbum. Heidelberg: Winter. Beringer, Nicole. ‘The use of German ‘braucht’ in unscripted speech: morphological reduction or phonetically caused Elision’. Bhatt, Rajesh and Roumyana Pancheva (2006). ‘Conditionals’. In: The Blackwell Companion to Syntax. Ed. by Martin Everaert and Henk R. van Riemsdijk. Oxford: Blackwell, pp. 638– 687. Bildhauer, Felix (2011). ‘Mehrfache Vorfeldbesetzung und Informationsstruktur: eine Bestandsaufnahme’. In: Deutsche Sprache 11.4, pp. 362–372. Birkmann, Thomas (1987). Präteritopräsentia: morphologische Entwicklung einer Sonderklasse in den altgermanischen Sprachen. Vol. 188. Linguistische Arbeiten. Tübingen: Niemeyer. Blatz, Friedrich (1880). Neuhochdeutsche Grammatik mit Berücksichtigung der historischen Entwicklung der deutschen Sprache. 2nd ed. Tauberbischofsheim: Lang. Blatz, Friedrich (1896). Neuhochdeutsche Grammatik mit Berücksichtigung der historischen Entwicklung der deutschen Sprache. 3rd ed. Vol. 2. Karlsruhe: Lang. Blatz, Friedrich (1900). Neuhochdeutsche Grammatik mit Berücksichtigung der historischen Entwicklung der deutschen Sprache. 3rd ed. Vol. 2. Karlsruhe: Lang. Bloch, Oscar and Walther von Wartburg (1986). Dictionnaire étymologique de la lange francaise. 7th ed. Paris: Presses Universitaires de France. Blühdorn, Hardarik (2012). Negation im Deutschen. Syntax, Informationsstruktur, Semantik. Vol. 48. Studien zur deutschen Sprache. Tübingen: Narr. Bödiker, Johann (1698). Grundsätze der deutschen Sprachen im Reden und Schreiben: samt einen Bericht vom rechten Gebrauch der Vorwörter, der studierenden Jugend und allen Deutschliebenden zum Besten vorgestellt. Berlin: Meyers Erben u. Zimmermann. Bolkestein, Machtelt A. (1980). Problems in the description of modal verbs: an investigation of Latin. Assen: Van Gorcum. Boye, Kasper (2010). ‘Raising verbs and auxiliaries in a functional theory of grammatical status’. In: Language usage and language structure. Ed. by Kasper Boye and Elisabeth Engberg-Pedersen. Berlin: Mouton de Gruyter. Braune, Wilhelm (1886). Althochdeutsche Grammatik. Halle: Niemeyer. Braune, Wilhelm and Ingo Reiffenstein (2004). Althochdeutsche Grammatik. Laut und Formenlehre. 15th ed. Tübingen: Niemeyer. Brennan, Virginia (1993). ‘Root and Epistemic Modal Auxiliary Verbs in English’. PhD thesis. Amherst: University of Massachusetts. Brennan, Virginia (1997). ‘Quantificational Modals’. In: Linguistic Inquiry 1.28, pp. 165–170. Bres, Jacques and Emanuelle Labeau (2012). ‘De la grammaticalisation des formes itive (‘aller’) et ventive (‘venir’): valeur en langue, emplois en discours’. In: Etudes de sémantique et pragmatique françaises. Ed. by Louis de Saussure and Alain Rihs. Bern: Lang, pp. 143– 166.

562 | References

Brinkmann, Hennig (1962). Die deutsche Sprache. Gestalt und Leistung. Düsseldorf: Schwann. Brünner, Gisela (1979). ‘Modales ’nicht-brauchen’ und ’nicht-müssen’’. In: Linguistische Berichte 62, pp. 81–93. Brünner, Gisela and Angelika Redder (1983). Studien zur Verwendung der Modalverben. Studien zur deutschen Grammatik 19. Tübingen: Narr. Bühler, Karl (1934). Sprachtheorie. Jena: Fischer. Buscha, Joachim, Gertraud Heinrich and Irene Zoch (1971). Modalverben. Leipzig: Enzyklopaedie Verglag. Butler, Jonny (2003). ‘A minimalist treatment of modality’. In: Lingua 113, pp. 967–996. Butt, Miriam (2006). Theories of Case. Cambridge: University Press. Bybee, Joan, Revere Dale Perkins and William Pagliuca (1994). The evolution of grammar. Chicago: University of Chicago Press. Calbert, Joseph P (1975). ‘Toward the Semantics of Modality’. In: Aspekte der Modalität. Ed. by Joseph P. Calbert and Heinz Vater. Vol. 1. Studien zur deutschen Grammatik. Narr, pp. 1– 70. Carlson, Gregory N. (1977). ‘Reference to kinds in English’. PhD thesis. Amherst: University of Massachusetts. Carpenter, Bob (1998). Type logical semantics. Cambridge, Mass: MIT press. Chomsky, Noam (1981). Lectures on government and binding: the Pisa lectures. Studies in generative grammar 9. Berlin: Mouton de Gruyter. Cinque, Guglielmo (1999). Adverbs and functional heads: a cross-linguistic perspective. Oxford studies in comparative syntax. New York: Oxford University Press. Claius, Johannes (1578). Grammatica Germanicae Lingvae ex bibliis Lutheris germanicis et aliis eius libris colecta. Leipzig. Clément, Danièle (1986). ‘A propos de faux vides et de faux PRO: les sujets des infinitives del’allemand introduites par ‘um’-‘zu’ sont-ils controlés?’ In: Revue de linguistique 34– 35, pp. 257–273. Clément, Danièle and Wolfgang Thümmel (1975). Grundzüge einer Syntax der deutschen Standartsprache. Frankfurt/Main: Athenäum Fischer. Coates, Jennifer (1983). The semantics of modal auxiliaries. London: Croom Helm. Cohen, Ariel (2010). Epistemic modals as speech act modifiers. Talk given at the ZAS Berlin, November 2010. Colomo, Katarina (2011). Modalität im Verbalkomplex. Halbmodalverben und Modalitätsverben im System statusregierender Verbklassen. Vol. 6. Bochumer Linguistische Arbeitsberichte. Bochum: Ruhr Universität Bochum. Comrie, Bernard (1989). ‘On identifying future tenses’. In: Tempus – Modus – Aspekt. Die lexikalischen und grammatischen Formen in den germanischen Sprachen. Ed. by Werner Abraham. Vol. 237. Linguistische Arbeiten. Tübingen: Niemeyer, pp. 51–64. Condoravdi, Cleo (2002). ‘Temporal interpretationof modals: modals for the present and modals for the past’. In: The construction of meaning. Ed. by David Beaver et al. Stanford: CSLI Publications, pp. 59–87. Coniglio, Marco (2008). ‘Die Syntax der deutschen Modalpartikeln: ihre Distribution und Lizenzierung in Haupt und Nebensätzen’. PhD thesis. Università Ca’ Foscari Venezia. Copley, Bridget (2006). ‘What should ’should’ mean?’ CNRS/Université Paris VIII. Cornillie, Bert (2007). Evidentiality and epistemic modality in Spanish (semi-)auxiliaries: a cognitive-functional approach. Vol. 5. Application of cognitive linguistics. Berlin: de Gruyter.

References |

563

Cornillie, Bert et al. (2009). ‘Modals in the Romance Languages’. In: Modals in the languages of Europe. Ed. by Björn Hansen and Ferdinand de Haan. Vol. 44. Empirical Approaches to Language Typology. Berlin: De Gruyter, pp. 107–138. Curme, George O. (1922). A grammar of the German language. New York: Frederick Ungar Publishing CO. Curme, George O. (1931). A grammar of the English language. Vol. 2 Syntax. Dal, Ingerid (1952). Kurze deutsche Syntax. Tübingen: Niemeyer. De Haan, Ferdinand (1997). The interaction of modality and negation: a typlogicial study. New York: Garland. Declerck, Renaat and Susan Reed (2001). Conditionals. A Comprehensive empirical analysis. Vol. 37. Topics in English Linguistics. Mouton de Gruyter. Demirdache, Hamida and Myriam Uribe-Etxebarria (2008). ‘Scope and anaphora with time arguments: The case of ’perfect modals’’. In: Lingua 118, pp. 1790–1815. Demske, Ulrike (2001). ‘Zur Distribution von Infinitivkomplementen im Althochdeutschen’. In: Modalität und Modalverben im Deutschen. Ed. by Marga Reis and Reimar Müller. Hamburg: Buske, pp. 61–85. Denison, David (1993). English Historical Syntax. London: Longman. Depraetere, Ilse and An Verhulst (2008). ‘Source of modality: a reassessment’. In: English Language and Linguistics 12.1, pp. 1–25. Diesing, Molly (1992). Indefinites. Linguistic Inquiry Monographs 20. Cambridge, MA: MIT Press. Dietrich, Rainer (1992). Modalität im Deutschen. Zur Theorie der relativen Modalität. Opladen: Westdeutscher Verlag. Diewald, Gabriele (1999). Die Modalverben im Deutschen: Grammatikalisierung und Polyfunktionalität. Reihe Germanistische Linguistik 208. Tübingen: Niemeyer. Diewald, Gabriele (2001). ‘‘Scheinen’-Probleme: Analogie, Konstruktionsmischung und die Sogwirkung aktiver Grammatikalisierungskanäle’. In: Modalität und Modalverben im Deutschen. Ed. by Reimar Müller and Marga Reis. Hamburg: Buske, pp. 87–110. Diewald, Gabriele and Elena Smirnova (2010). Evidentiality in German: linguistic realization and regularities in grammaticalization. Vol. 228. Trends in linguistics. Berlin: Mouton de Gruyter. Dik, Simon C. (1975). ‘The semantic representation of manner adverbials’. In: Linguistics in the Netherlands 1972–1973. Ed. by A Kraak. Assen: Van Gorcum, pp. 96–121. Doherty, Monika (1985). Epistemische Bedeutung. Vol. 23. Studia grammatica. Berlin: Akademie Verlag. Doitchinov, Serge (2001). ‘’Es kann sein, daß der Junge ins Hausgegangen ist’ - Zum Erstspracherwerb von ’können’ in epistemischer Lesart.’ In: Modalität und Modalverben im Deutschen. Ed. by Reimar Müller and Marga Reis. Hamburg: Buske, pp. 111–134. Dougherty, Ray C. (1970). ‘A grammar of coördinate conjoined structures I’. In: Language 46.4, pp. 850–898. Dowty, David (1991). ‘Thematic proto-roles and argument selection’. In: Language 67, pp. 547– 619. Drosdowski, Günther et al. (1984). Duden: Grammatik der deutschen Gegenwartssprache. 4th ed. Vol. 4. Der Duden. kAb 1800/4/4 ==> (Phil). Mannheim: Dudenverlag. Drosdowski, Günther et al. (1995). Duden: Grammatik der deutschen Gegenwartssprache. 5th ed. Vol. 4. Der Duden. Mannheim: Dudenverlag.

564 | References

Drubig, Hans Bernhard (2001). ‘On the syntactic form of epistemic modality’. Ms. Universität Tübingen. Ebert, Robert Peter (1976). Infinitival Constructions in Early New High German. Linguistische Arbeiten 30. Tübingen: Niemeyer. Ebert, Robert Peter et al. (1993). Frühneuhochdeutsche Grammatik. Tübingen: Niemeyer. Ehlich, Konrand and Jochen Rehbein (1972). ‘Einige Interrelationen von Modalverben’. In: Linguistische Pragmatik. Ed. by Dieter Wunderlich. Frankfurt am Main: Athenäum, pp. 318– 340. Ehrich, Veronika (2001). ‘Was ’nicht müssen’ und ’nicht können’ (nicht) bedeuten können: zum Skopus der Negation bei den Modalverben des Deutschen.’ In: Modalität und Modalverben im Deutschen. Ed. by Marga Reis and Reimar Müller. Hamburg: Buske, pp. 149–176. Eide, Kristin Melum (2005). Norwegian Modals. Studies in Generative Grammar 74. Berlin: Mouton de Gruyter. Eisenberg, Peter (1992). ‘Adverbiale Infinitive: Abgrenzung, Grammatikalisierung, Bedeutung’. In: Deutsche Syntax. Ansichten und Aussichten. Ed. by Ludger Hoffmann. Berlin, pp. 206– 224. Eisenberg, Peter (2004). Grundriß der deutschen Grammatik. 2. Auflage. Vol. 2. Stuttgart: Metzler. Eisenberg, Peter et al. (2005). Duden, die Grammatik: unentbehrlich für richtiges Deutsch. Ed. by Kathrin Kunkel-Razum. 7th ed. Vol. 4. Der Duden. Mannheim: Dudenverlag. Enç, Mürvet (1996). ‘Tense and Modality’. In: The Handbook of Contemporary Semantic Theory. Ed. by Shalom Lappin. Oxford: Blackwell, pp. 345–358. Engel, Ulrich (1991). Deutsche Grammatik. 2. verbesserte Auflage. Heidelberg: Groos. Engel, Ulrich (1996). Deutsche Grammatik. 3rd ed. Heidelberg: Groos. Erb, Marie Christine (2001). ‘Finite Auxiliaries in German’. PhD thesis. Katholieke Universiteit Brabant. Eroms, Hans-Werner (1980). ‘Funktionskonstanz und Systemstabilisierung bei den begründenden Konjunktionen im Deutschen’. In: Sprachwissenschaft 5, pp. 73–115. Fabricius-Hansen, Cathrine (1980). ‘Sogenannte ergänzende ‘wenn’-Sätze. Ein Beispiel zu syntaktisch- semantischer Argumentation’. In: Festschrift für Gunnar Bech zum 60. Geburtstag, pp. 160–188. Fabricius-Hansen, Cathrine (1999). ‘‘Moody Time’: Indikativ und Konjunktiv im deutschen Tempussystem’. In: Zeitschrift für Sprachwissenschaft und Linguistik 29.13, pp. 119–146. Fabricius-Hansen, Cathrine (2000). ‘Die Geheimnisse der deutschen ‘würde’-Konstruktion’. In: Deutsche Grammatik in Theorie und Praxis. Ed. by Rolf Thieroff. Tübingen, pp. 183–196. Fabricius-Hansen, Cathrine and Kjell Johan Sæbø (1983). ‘Über das Chamäleon ‘wenn’ und seine Umwelt’. In: Linguistische Berichte 83, pp. 1–35. Fagan, Sarah M. B. (2001). ‘Epistemic Modality and Tense in German’. In: Journal of Germanic Linguistics 13.3, pp. 197–230. Faller, Martina (2010). ‘A possible worlds semantics for Cuzco Quechua evidentials’. In: Proceedings of SALT 20, 29 Apr 2010-01 May 2010, Vancouver. Ed. by Nan Li and David Lutz, pp. 660–683. Faller, Martina (2011). Reportative evidentials and modal subordination. Talk given at the CHRONOS 2011, Aston University Birmingham. Faller, Martina (2012). ‘Evidential scalar implicatures’. In: Linguistics and Philosophy 35, pp. 285–312.

References |

565

Feldman, Fred (1986). Doing the best we can: an essay in informal deontic logic. Vol. 35. Philosophical studies series in philosophy. Dordrecht: Reidel. Fillmore, Charles J. (1997). ‘Lectures on deixis’. In: Stanford, Calif.: Center for the Study of Language and Information. Chap. Coming and Going, pp. 77–102. Fintel, Kai von and Anthony S. Gillies (2008). ‘CIA Leaks’. In: Philosophical Review 1.117, pp. 77– 98. Fintel, Kai von and Anthony S. Gillies (2010). ‘Must...Stay...Strong!’ In: Natural Language Semantics 18.4, pp. 351–383. Fintel, Kai von and Anthony S. Gillies (2011). ‘’Might’ made right’. In: Epistemic modality. Ed. by Andy Egan and Brian Weatherson. Oxford: Oxford University Press, pp. 108–130. Fintel, Kai von and Sabine Iatridou (2003). ‘Epistemic containment’. In: Linguistic Inquiry 2.34, pp. 173–198. Fleischer, Jürg and Oliver Schallert (2011). Historische Syntax des Deutschen. Narr Studienbücher. Tübingen: Narr. Folsom, Marvin H. (1968). ‘’Brauchen’ im System der Modalverben’. In: Muttersprache 78, pp. 321–329. Fourquet, Jean (1970). ‘Zum subjektiven Gebrauch der deutschen Modalverben’. In: Studien zur Syntax des heutigen Deutsch, Paul Grebe zum 60. Geburtstag. Vol. VI. Sprache der Gegenwart. Düsseldorf: Schwann, pp. 154–161. Freywald, Ulrike (2010). ‘‘Obwohl vielleicht war es ganz anders’. Vorüberlegungen zum Alter der Verbzweitstellung nach subordinierenden Konjunktionen’. In: Historische Texgrammatik und historische Syntax des Deutschen. Ed. by Arne Ziegler. Berlin: De Gruyter, pp. 54–84. Fritz, Gerd (1991). ‘Deutsche Modalverben 1609 - epistemische Verwendungsweisen’. In: Beiträge zur Geschichte der deutschen Sprache und Literatur 113, pp. 28–53. Fritz, Gerd (1997). ‘Historische Semantik der Modalverben’. In: Untersuchungen zur semantischen Entwicklungsgeschichte der Modalverben im Deutschen. Ed. by Gerd Fritz and Thomas Gloning. Reihe Germanistische Linguistik 187. Tübingen: Niemeyer, pp. 1– 157. Gamon, David (1993). ‘On the development of epistemicity in the German modalverbs ‘mögen’ and ‘müssen’’. In: Folia Linguistica Historica 14, pp. 125–176. Gazdar, Gerald (1981). ‘Unbounded Dependencies and Coordinate Structure’. In: Linguistic Inquiry 12, pp. 155–184. Geilfuß, Jochen (1992). ‘Ist ’wollen’ ein Kontrollverb oder nicht?’ In: Arbeitspapiere des SFB 340. Gelhaus, Hermann (1969). ‘Strukturanalyse und Statistik - Über den Widerstreit zweier Kriterien’. In: Wirkendes Wort 19, pp. 310–324. Gergel, Remus and Jutta Hartmann (2009). ‘Experiencers with (un)-willingness’. In: Advances in comparative germanic syntax. Ed. by Artemis Alexiadou et al. Vol. 141. Linguistik Aktuell. Amsterdam: John Benjamins, pp. 327–355. Geuder, Wilhelm (2006). ‘Manner modification of states’. In: Proceedings of Sinn und Bedeutung. Ed. by Christian Ebert and Cornelia Endriss. ZAS Papers in Linguistics. Berlin: ZAS, pp. 111–124. Ginzburg, Jonathan (2012). The interactive stance: meaning for conversation. Oxford University Press. Girnth, Heiko (2000). Untersuchungen zur Theorie der Grammatikalisierung am Beispiel des Westmitteldeutschen. Vol. 223. Reihe germanistische Linguistik. Tübingen: Niemeyer.

566 | References

Glas, Reinhold (1984). ‘sollen’ im heutigen Deutsch. Bedeutung und Gebrauch in der Schriftsprache. Studien zur deutschen Grammatik 27. Tübingen: Narr. Götzke, Alfred and Walther Mitka (1939). Trübners deutsches Wörterbuch. A-B. Vol. 1. Berlin: de Gruyter. Grebe, Paul (1959). Duden: Grammatik der deutschen Gegenwartssprache. 1st ed. Vol. 4. Der große Duden. Mannheim: Dudenverlag. Grebe, Paul et al. (1966). Duden: Grammatik der deutschen Gegenwartssprache. 2nd ed. Vol. 4. Der große Duden. dk 1236-4 ==> (Phil), Spr 7 b-dud 2.A. b ==> (ErzPsy). Mannheim: Dudenverlag. Grebe, Paul et al. (1973). Duden: Grammatik der deutschen Gegenwartssprache. 3rd ed. Vol. 4. Der große Duden. spr 7 b-dud 3.A. ==> (ErzPsy). Mannheim: Dudenverlag. Grice, Paul (1989). Studies in the way of words. Cambridge, Mass.: Harvard University Press. Gries, Stefan Th. and Anatol Stefanowitsch (2004). ‘Extending collostructional analysis’. In: International Jounral of Corpus Linguistics 9.1, pp. 97–129. Griesbach, Heinz and Dora Schulz (1960). Grammatik der deutschen Sprache. Hueber. Griesbach, Heinz and Dora Schulz (1976). Grammatik der deutschen Sprache. 10th ed. Hueber. Grimm, Jacob (1822). Deutsche Grammatik I. Göttingen: Dieterich. Grimm, Jacob (1837). Deutsche Grammatik IV. Göttingen: Dietrich. Grimm, Jacob and Wilhelm Grimm (1860). Deutsches Wörterbuch. Vol. 2. Leipzig: Hirzel. Grosz, Patrick (2014). ‘Modal particles in rationale clauses and related constructions’. In: Modes of Modality. Ed. by Elisabeth Leiss and Werner Abraham. Vol. 149. Studies in language companion series. Amsterdam: John Benjamins, pp. 263–290. Gunlogson, Christine (2001). ‘True to form: rising and falling declaratives as questions in English’. PhD thesis. Santa Cruz: University of California. Günthner, Susanne (1996). ‘From subordination to coordination?: verb second position in German causal and concessive constructions’. In: Pragmatics 6.3, pp. 323–370. Günthner, Susanne (1999). ‘Entwickelt sich der Konzessivkonnektor obwohl zum Diskursmarker? Grammatikalisierungstendenzen im gesprochenen Deutsch.’ In: Linguistische Berichte 180, pp. 409–446. Günthner, Susanne (2000a). ‘From concessive connector to discourse marker: The use of ‘obwohl’ in everyday German Interaction’. In: Cause, Condition, Concession, Contrast. Cognitive and Discourse Perspectives. Ed. by Elizabeth Couper-Kuhlen and Bernd Kortmann. Berlin: de Gruyter, pp. 439–486. Günthner, Susanne (2000b). ‘’wobei (.) es hat alles immer zwei seiten.’ Zur Verwendung von ’wobei’ im gesprochenen Deutsch’. In: Deutsche Sprache 28.4, pp. 313–341. Günthner, Susanne (2002). ‘Zum kausalen und konzessiven Gebrauch des Konnektors ’wo’ im gesprochenen Umgangsdeutsch’. In: 30.3, pp. 310–341. Günthner, Susanne (2008). ‘’weil - es ist zu spät’ Geht die Nebensatzstellung im Deutschen verloren?’ In: Frischwärts und Unkaputtbar. Sprachverfall und Sprachwandel im Deutschen? Ed. by Markus Denkler, Susanne Günthner and Wolfgang Imo. Aschendorff, pp. 103–128. Hacquard, Valentine (2006). ‘Aspects of Modality’. PhD thesis. MIT. Hacquard, Valentine (2010). ‘On the event relativity of modal auxiliaries’. In: Natural Language Semantics 18, pp. 17–114. Hacquard, Valentine and Alexis Wellwood (2012). ‘Embedding epistemic modals in English. A corpus-based study’. In: Semantics & Pragmatics 5.4, pp. 1–29. Haegeman, Liliane (2002). ‘Anchoring to speaker, adverbial clauses and the structure of CP’. In: Georgetown University Working Papers in Theoretical Linguistics 2, pp. 117–180.

References |

567

Haegeman, Liliane (2004). ‘Topicalization, CLLD and the Left Periphery’. In: ZAS Papers in Linguistics 35, pp. 157–192. Haegeman, Liliane (2006). ‘Conditionals, factives and the left periphery’. In: Lingua 116, pp. 1651–1669. Haider, Hubert (1993). Deutsche Syntax – generativ. Vol. 325. Tübinger Beiträge zur Linguistik. Tübingen: Narr. Haider, Hubert (2010). The Syntax of German. Cambridge Sytax Guides. Cambridge: Cambridge University Press. Hansen, Björn and Ferdinand De Haan (2009). Modals in the languages of Europe: a reference work. Vol. 44. Empirical Approaches to Language Typology. Berlin: Mouton de Gruyter. Heim, Irene (1982). ‘The semantics of definite and indefinite noun phrases’. PhD thesis. Heine, Bernd (1995). ‘Agent-oriented vs. Epistemic Modality. Some Observations on German Modals’. In: Modality in Grammar and Discourse. Ed. by Joan Bybee and Suzanne Fleischmann. Vol. 32. Typological Studies in Language. John Benjamins, pp. 17–53. Helbig, Gerhard and Joachim Buscha (2001). Deutsche Grammatik: ein Handbuch für den Ausländerunterricht. Berlin: Langenscheidt. Helbig, Gerhard and Agnes Helbig (1990). Lexikon deutscher Modalwörter. Leipzig: Enzyklopädie Verlag. Hengeveld, Kees (1988). ‘Illocution, mood and modality in a functional grammar of Spanish’. In: Journal of Semantics 6, pp. 227–269. Hetland, Jorunn and Heinz Vater (2008). ‘Zur Syntax der Modalverben im Deutschen und Norwegischen’. In: Modalität und Grammatikalisierung. Ed. by Eva Maagerø, Ole Letnes and Heinz Vater. Vol. 34. Fokus. Trier: Wissenschaftlicher Verlag, pp. 91–116. Heyse, Johann Christian August (1822). Theoretisch=praktische deutsche Grammatik oder Lehrbuch zum reinen und richtigen Sprechen, Lesen und Schreiben der deutschen Sprache. 3rd ed. Hannover: Hahn’sche Hof=Buchhandlung. Hill, Virginia (2011). ‘Modal grammaticalization and the pragmatic field’. In: Diachronica 28.1, pp. 25–53. Hinterhölzl, Roland (2009). ‘The IPP-effect, phrasal affixes and repair strategies in the syntax morphology interface’. In: Linguistische Berichte 218, pp. 191–216. Hinterwimmer, Stefan (2014). The semantics of ‘sollen’. Talk given at Universität Tübingen. Hofmann, Thomas Ronald (1976). ‘Past Tense Replacement and the Modal System’. In: Notes from the Linguistic Underground. Ed. by James McCawley. Syntax and semantics 7. New York: Academic Press, pp. 86–99. Höhle, Tilman N. (1978). Lexikalistische Syntax. Die Aktiv-Passiv-Relation und andere Infinitivkonstruktionen im Deutschen. Vol. 67. Linguistische Arbeiten. Niemeyer. Höhle, Tilman N. (1982). ‘Explikation für “normale Betonung” und “normale Wortstellung”’. In: Satzglieder im Deutschen. Ed. by Werner Abraham. Studien zur deutschen Grammatik 15. Tübingen: Narr, pp. 75–153. Höhle, Tilmann (1992). ‘Über Verum-Fokus im Deutschen’. In: Informationsstruktur und Grammatik. Ed. by Joachim Jacobs. Linguistische Berichte Sonderheft 4/1991–1992. Opladen: Westdeutscher Verlag, pp. 112–141. Holler, Anke (2009). ‘Towards an analysis of the adverbial use of German interrogative ‘was’ (‘what’)’. In: Proceedings of the HPSG09 Conference, Georg-August-Universität Göttingen. Ed. by Stefan Müller. 131–149. Stanford: CSLI Publications. Homer, Vincent (2010). ‘Epistemic Modals: High ma non troppo’. In: Proceedings of the 40th Meeting of the North East Linguistic Society. Amherst, MA.

568 | References

Hornstein, Norbert (2003). ‘On Control’. In: Minimalist Syntax. Ed. by Randall Hendricks. Malden: Blackwell, pp. 6–81. Huitink, Janneke (2008). ‘Scoping over epistemics in English and in Dutch’. In: Collection of the papers selected from the CIL held at Korea University in Seoul, Linguistic Society of Korea, pp. 2077–2089. Jackendoff, Ray (1972). Semantic interpretation in generative grammar. Vol. 2. Studies in linguistics series. Cambridge, MA: MIT Press. ¯ Jackendoff, Ray (1977). X-syntax: a study of phrase structure. Camebridge, Mass: MIT Press. Jacobs, Joachim (1988). ‘Fokus-Hintergrund-Gliederung und Grammatik’. In: Intonationsforschungen. Ed. by Hans Altmann. Linguistische Arbeiten 200. Niemeyer, pp. 89–134. Jäger, Agnes (2010). ‘Der Komparativzyklus und die Position der Vergleichspartikeln’. In: Linguistische Berichte 224, pp. 467–493. Jäger, Gerhard (2001). ‘Topic-comment structure and the contrast between stage level and individual level predicates’. In: Journal of Semantics 18, pp. 83–126. Jäger, Siegfried (1968). ‘Ist ’ brauchen’ mit ’zu’ nicht sprachgerecht?’ In: Muttersprache 78, pp. 330–333. Jellinek, Max Hermann (1914). Geschichte der neuhochdeutschen Grammatik von den Anfängen bis auf Adelung. Heidelberg: Winter. Jespersen, Otto (1917). Negation in English and other languages. Vol. 1. Historisk-filologiske meddelser 5. Hoest. Jespersen, Otto (1931). A Modern English Grammar. Syntax. Third volume. Time and Tense. Heidelberg: Carl Winter. Jędrzejowski, Łukasz (2010). ‘Zur epistemischen Infinitheitslücke aus diachroner und typologischer Sicht’. MA thesis. Freie Universität Berlin. Johnen, Thomas (2003). Die Modalverben des Portugiesischen (PB und PE). Semantik und Pragmatik in der Verortung einer kommunikativen Grammatik. Hamburg: Dr. Kovac. Johnen, Thomas (2006). ‘Zur Herausbildung der Kategorie ’Modalverb’ in der Grammatikographie des Deutschen’. In: Pandaemonium germanicum 10, pp. 283–338. Kaeding, Friedrich Wilhelm (1897). Die Hilfszeitwörter in ihrem Verhältnis zum deutschen Wortschatz. Berlin Steglitz: Selbstverlag. Kaiaty, Mohamed (2010). ‘Überlegungen zu sog. ‘ergänzenden ‘wenn’-Sätzen’ im Deutschen’. In: Deutsche Sprache 38.4, pp. 287–308. Kaita, Kousuke (2012). ‘Modal auxiliaries from Late Old to Early Middle English’. PhD thesis. München: Ludwig Maximilians Universität. Kasper, Walter (1987). Semantik des Konjunktivs II in Deklarativsätzen des Deutschen. Vol. 71. Reihe germanistische Linguistik. Tübingen: Niemeyer. Kiaulehn, Walther (1965). Der richtige Berliner in Wörtern und Redensarten. Ed. by Hans Meyer. 10th ed. München: Biederstein. Kiefer, Ferenc (1984). ‘Focus and modality’. In: Groninger Abreiten zur Germanistischen Linguistik 24, pp. 55–81. Kirchner, Gustav (1940). ‘’to want’ as an auxiliary of modality’. In: English Studies 22.1-6, pp. 129–136. Kiss, Tibor (1995). Infinitive Komplementation. Neue Studien zum deutschen Verbum infinitum. Linguistische Arbeiten 333. Niemeyer. Kissine, Mikhail (2008). ‘Why ’will’ is not a modal.’ In: Natural Language Semantics 16, pp. 129–155. Klein, Wolfgang (1994). Time in Language. London: Routledge.

References |

569

Kluempers, John David (1997). The grammaticalization of a verb: the role of ’nicht brauchen’ in the German modal verb system. Ann Arbor: UMI. Kluge, Friedrich (2011). Etymologisches Wörterbuch der deutschen Sprache. 25th ed. Berlin: de Gruyter. Kolb, Hermann (1964). ‘Über ’brauchen’ als Modalverb’. In: Zeitschrift für deutsche Sprache 20, pp. 64–78. König, Ekkehard and Johan van der Auwera (1988). ‘Clause integration in German and Duthc conditionals, concessive conditionals and concessives’. In: Clause combining in grammar and discourse. Ed. by John Haiman and Sandra A. Thompson. Amsterdam: John Benjamins, pp. 101–133. Kotin, Michail L. (2008). ‘Aspects of a reconstruction of form and function of modal verbs’. In: Modality-aspect interfaces. Ed. by Werner Abraham and Elisabeth Leiss. Vol. 79. Typological Studies in Language. Amsterdam: John Benjamins, pp. 371–384. Kramer, Matthias (1702a). Das herrlich-grosse Teutsch-Italiänische Dictionarium. Nürnberg. Krämer, Sabine (2005). ‘Synchrone Analyse als Fenster zur Diachronie. Die Grammatikalisierung von ’werden’ + Infinitiv’. PhD thesis. Humboldt Universität zu Berlin. Kratzer, Angelika (1978). Semantik der Rede. Kontexttheorie - Modalwörter - Konditionalsätze. Scriptor. Kratzer, Angelika (1981). ‘The Notional Category of Modality’. In: Words, Worlds and Contexts. New approaches in World Semantics. Ed. by Hans Jürgen Eikmeyer and Hannes Rieser. Berlin: Mouton de Gruyter, pp. 38–74. Kratzer, Angelika (1986). ‘Conditionals’. In: Papers from the parasession on pragmatics and grammatical theory at the 22nd regional meeting. Ed. by A. M. Farley and P. T. Farley. Chicago Linguistic Society, pp. 1–15. Kratzer, Angelika (1991). ‘Modality’. In: Semantik. Ein internationales Handbuch der zeitgenössischen Forschung. Ed. by Arnim von Stechow and Dieter Wunderlich. 639–650. Berlin: Mouton de Gruyter. Kratzer, Angelika (1995). ‘Stage-level and individual-level predicates’. In: The Generic book. Ed. by Gregory N. Carlson and F. J. Pelletier. Chicago: University of Chicago Press, pp. 125–175. Kratzer, Angelika (2011). Epistemic modals, embedded, modified, and plain. Talk given at the 33rd DGfS, Göttingen. Kratzer, Angelika (2012). Modals and Conditionals. Oxford Studies in Theoretical Linguistics. Oxford: Oxford University Press. Krause, Maxi (1997). ‘Zur Modalisierung bei Otfrid’. In: Semantik der syntaktischen Beziehungen. Akten des Pariser Kolloquiums zur Erforschung des Althochdeutschen 1994. Ed. by Yvon Desportes. Vol. 3. Germanistische Bibliothek. Heidelberg: Winter, pp. 92–106. Krifka, Manfred (2007). ‘Basic notions of information structure’. In: Interdisciplinary studies of information structure. Ed. by Caroline Fery and Manfred Krifka. Vol. 6. Potsdam. Kurrelmeyer, Wilhelm (1910). ‘Über die Entstehung der Konstruktion ’Ich habe sagen hören’’. In: Zeitschrift für deutsche Wortforschung 12, pp. 157–173. Kürschner, Wilfried (1983). Studien zur Negation im Deutschen. Studien zur deutschen Grammatik 12. Tübingen: Narr. Lang, Ewald (1979). ‘Zum Status der Satzadverbiale’. In: Slovo a Slovesnost 40, pp. 200–213. Langer, Nils (2001). Linguistic Purism in Action. How auxiliary ’tun’ was stigmatized in Early New High German. Berlin: de Gruyter. Lasersohn, Peter (2005). ‘Context dependence, disagreement and predicates of personal taste’. In: Linguistics and Philosophy 28, pp. 643–686.

570 | References

Lassiter, Daniel (2011). ‘Measurement and modality: the scalar basis of modal semantics’. PhD thesis. New York University. Lechner, Winfried (2006). ‘An Interpretive effect of head movement’. In: Phases of Interpretation. Ed. by Mara Frascarelli. Berlin: Mouton de Gruyter. Leech, Geoffrey N. (1971). Meaning and the English verb. London: Longman. Lehmann, Christian (1995). Thoughts on Grammaticalization. Revised and expanded version. Vol. 1. LINCOM Studies in Theoretical Linguistics. Lincom Europa. Leiss, Elisabeth (2002). ‘Explizite und implizite Kodierung von Deontizität und Epistemizität’. In: Jezykoslovlje 3, pp. 69–98. Lenz, Barbara (1996). ‘Wie ’brauchen’ ins deutsche Modalverb-System geriet und welche Rolle es darin spielt.’ In: Beiträge zur Geschichte der deutschen Sprache und Literatur 118, pp. 393–422. Letnes, Ole (2002). ‘‘Wollen’ zwischen Referat und Verstellung’. In: Modus, Modalverben, Modalpartikel. Ed. by Cathrine Fabricius-Hansen, Oddleif Leirbukt and Ole Letnes. LinguistischPhilologische Studien 25. Trier: Wissenschaftlicher Verlag Trier, pp. 101–120. Levinson, Stephen (1983). Pragmatics. Camebridge textbooks in linguistics. Camebridge: Camebridge University Press. Levinson, Stephen C. (2000). Presumptive meanings. Cambridge, Mass.: MIT Press. Lewis, David (1979). ‘Scorekeeping in a language game’. In: Semantics from different points of view. Ed. by Rainer Bäuerle, Urs Egli and Arnim von Stechow. Berlin: Springer, pp. 172– 187. Leys, Odo (1991). ‘Skizze einer kognitiv-semantischen Typologie der deutschen ’um’-Infinitive’. In: Leuvense Bijdragen 80, pp. 167–203. Lightfoot, David (1979). Principles of diachronic syntax. Vol. 23. Cambridge studies in linguistics. Cambridge: Cambridge University Press. Lima, José Pinto de (2004). ‘On the evidential verbs in German and Portuguese: the grammaticalization of ’scheinen’ and ’parecer’’. In: Linguistic studies in the European Year of Languages. Proceedings of the 36th Linguistic Colloquium, Ljubljana 2001. Ed. by Stojan Bračič et al. Frankfurt/Main: Peter Lang, pp. 415–424. Lima, José Pinto de (2005). ‘Zur Grammatikalisierung von dt. ’drohen’ und pg. ’ameaçar’’. López, Luis and Susanne Winkler (2000). ‘Focus and topic in VP-anaphora constructions’. In: Linguistics 38.4, pp. 623–664. Lötscher, Andreas (1991). ‘Der Konjunktiv II bei Modalverben und die Semantik des Konjunktiv II’. In: Sprachwissenschaft 16, pp. 334–364. Lyons, John (1977). Semantics. Vol. 2. Cambridge: Cambridge University Press. Maché, Jakob (2008). ‘The autopsy of a modal - insights from the historical development of German’. In: Modality-Aspect Interfaces. Implications and Typological Solutions. Ed. by Werner Abraham and Elisabeth Leiss. Vol. 79. Typological Studies in Language. Amsterdam: Benjamins, pp. 385–415. Maché, Jakob (2009). ‘Das Wesen epistemischer Modalität’. In: Modalität, Epistemik und Evidentialität bei Modalverb, Adverb, Modalpartikel und Modus. Ed. by Werner Abraham and Elisabeth Leiss. Vol. 77. Studien zur deutschen Grammatik. Tübingen: Stauffenburg, pp. 25–55. Maché, Jakob (2012). ‘Exploring the theory of mind interfaces’. In: Theory of Mind Elements across Languages. Ed. by Werner Abraham and Elisabeth Leiss. Trends in Linguistics 243. Berlin: de Gruyter, pp. 109–146.

References |

571

Maché, Jakob and Werner Abraham (2011). ‘Infinitivkomplemente im Frühen Neuhochdeutsch satzwertig oder nicht?’ In: Germanistische Linguistik. 235–274 213-215, pp. 235–274. Maienborn, Claudia (1994). ‘Kompakte Strukturen: Direktionale PPn und nicht-lokale Verben’. In: Kognitive Linguistik. Repräsentationen und Prozesse. Ed. by Sascha Felix, Christopher Habel and Gert Rickheit. Opladen: Westdeutscher Verlag, pp. 229–249. Maienborn, Claudia (2003). Die logische Form von Kopulasätzen. studia gramatica 56. Berlin: Akademie Verlag. Maienborn, Claudia (2004). ‘A pragmatic explanation of the stage level / individual level contrast in combination with locatives’. In: Proceedings of the WECOL. Ed. by Brian Agbayabi, Vida Samiian and Benjamnin Tucker. Vol. 15. Fresno: CSU, pp. 158–170. Maienborn, Claudia (2007). ‘Das Zustandspassiv. Grammatische Einordnung - Bildungsbeschränkungen - Interpretationsspielraum (erscheint)’. In: Zeitschrift für Germanistische Linguistik 35, pp. 83–115. Maitz, Péter and Krisztián Tronka (2009). ‘‘brauchen’ - Phonologische Aspekte der Auxiliarisierung’. In: Zeitschrift für Dialektologie und Linguistik LXXVI.2, pp. 189–202. Martin, Fabienne (2011). ‘Epistemic modals in the past’. McDowell, Joyce (1987). ‘Assertion and Modality’. PhD thesis. University of Southern California. Meid, Wolfgang (1971). Das germanische Präteritum. Indogermanische Grundlagen und Ausbreitung im Germanischen. Vol. 3. Innsbrucker Beiträge zur Sprachwissenschaft. Innbruck: Kowatsch. Menedéz-Benito, Paula (2010). ‘On universal free choice items’. In: Natural Language Semantics 18, pp. 33–64. Merchant, Jason (2009). ‘Ellipsis’. In: Syntax: an international handbook. Ed. by Artemis Alexiadou and Tibor Kiss. 2nd ed. Berlin: Mouton de Gruyter. Milan, Carlo (2001). Modalverben und Modalität: eine kontrastive Untersuchung Deutsch - Italienisch. Vol. 444. Linguistische Arbeiten. Niemeyer. Mortelmans, Tanja (2000). ‘Konjunktiv II and epistemic modal verbs in German’. In: Constructions in Cognitive Linguistics. Ed. by Ad Foolen and Frederike van der Leek. Vol. 178. Current Issues in Linguistic Theory. Amsterdam: John Benjamins, pp. 191–215. Mortelmans, Tanja, Kasper Boye and Johan van der Auwera (2009). ‘Modals in the Germanic languages’. In: Modals in the languages of Europe. Ed. by Björn Hansen and Ferdinand de Haan. Vol. 44. Empirical Approaches to Language Typology. Berlin: De Gruyter, pp. 11–69. Moscati, Vincenzo (2006). ‘The scope of negation’. PhD thesis. Università di Siena. Moulton, Keir (2010). ‘Small Clauses do reconstruct’. To be published in Proceedings of NELS 41. Müller, Reimar (2001). ‘Modalität im Prosalanzelot’. In: Modalität und Modalverben im Deutschen. Ed. by Marga Reis and Reimar Müller. Hamburg: Buske, pp. 239–262. Müller, Stefan (2002). Complex predicates: verbal complexes, resultative constructions and particle verbs in German. Stanford, CA: CSLI. Müller, Stefan (2007). Head-Driven Phrase Structure Grammar. Eine Einführung. Vol. 17. Stauffenburg Einführungen. Tübingen: Stauffenburg. Müller, Stefan (2009). ‘On predication’. In: Proceedings of the HPSG09 Confernence, University of Goettingen. Ed. by Stefan Müller. Stanford: CSLI publications, pp. 213–233. Müller, Stefan (2013). ‘On the Copula, Specificational Constructions and Type Shifting’. Murray, James A.H. (1933). A new English dictionary on historical principles. Vol. 7. Oxford: Clarendon.

572 | References

Neugeborn, Wolfgang (1976). ‘Zur Analyse von Sätzen mit finiter Verbform + Infinitiv’. In: Untersuchungen zur Verbvalenz. Ed. by H. Schumacher, pp. 66–74. Newmeyer, Frederick J. (1970). ‘The ’root modal’: can it be transitive?’ In: Studies presented to Robert B. Lees by his students. Ed. by Jerrold M. Sadock and Anthony Vanek. Vol. 1. Papers in linguistic monographs series. Edmonton: Linguistic Research, pp. 189–196. Nissenbaum, Jon (2005). ‘Kissing Pedro Martinez: (existential) anankastic conditionals and rationale clauses’. In: Proceedings of SALT XV. Cornell University. Ithaca, NY, pp. 134–151. Nuyts, Jan (2001a). Epistemic Modality, Language, and Conceptualization. Amsterdam: John Benjamins. Nuyts, Jan (2001b). ‘Subjectivity as an evidential dimension in epistemic modal expression’. In: Journal of Pragmatics 33.3, pp. 383–400. Öhlschläger, Günther (1989). Zur Syntax und Semantik der Modalverben. Vol. 144. Linguistische Arbeiten. Tübingen: Niemeyer. Ölinger, Albert (1574). Vnderricht der Hoch Teutschen Spraach. Straßburg. Ørsnes, Bjarne (2007). ‘Form und Funktion bei der Verwendung des Ersatzinfinitivs im Deutschen. Zum Gebrauch des Ersatzinfinitivs bei Fügung mit Modalverb und Direktionalergänzung’. In: Acta Linguistica Hafniensia 40, pp. 121–158. Pafel, Jürgen (1989). ‘’scheinen’ + Infinitiv. eine oberflächengrammatische Analyse’. In: Wissen, Wahrnehmung, Glauben: epistemische Ausdrücke und propositionale Einstellungen. Ed. by G. Falkenberg. Vol. 202. Linguistische Arbeiten. GELESEN. Niemeyer, pp. 123–172. Palmer, Frank R. (1990). Modality and the English modals. 2nd ed. London: Longman. Palmer, Frank Robert (1986). Mood and Modality. Cambridge text books in linguistics. Cambridge: Cambridge University Press. Papafragou, Anna (2006). ‘Epistemic modality and truth conditions’. In: Lingua 116, pp. 1688– 1702. Paul, Hermann (1897). Deutsches Wörterbuch. Halle a.S.: Niemeyer. Paul, Hermann (1898). Prinzipien der Sprachgeschichte. 3rd ed. Halle a. d. Salle: Niemeyer. Paul, Hermann (1917). Deutsche Grammatik. Vol. 2. Halle: Niemeyer. Paul, Hermann (1920). Deutsche Grammatik. Vol. 4 Syntax II. Niemeyer. Perkins, Michael R. (1983). Modal expressions in English. Norwood, New Jersey: Ablex Publishing Corporation. Perlmutter, David (1970). ‘The two verbs ’begin’’. In: Reading in English Transformational Grammar. Ed. by Roderick Jacobs and Peter S. Rosenbaum. Boston: Ginn, pp. 107–119. Pfeffer, Alan J. (1973). ‘’Brauchen’ als Vollverb, Hilfsverb und Modalverb’. In: Wirkendes Wort 23, pp. 86–92. Pietandrea, Paola (2005). Epistemic modality: functional properties and the Italian system. Studies in language companion series. Amsterdam: John Benjamins. Pillot, Jean (1550). Gallicae Linguae institutio. Paris. Pittner, Karin (2003). ‘Kasuskonflikte bei freien Relativsätzen’. In: Deutsche Sprache, pp. 193– 208. Pokorny, Julius (1959). Indogermanisches etymologisches Wörterbuch. Bern: Francke. Pollard, Carl and Ivan A. Sag (1994). Head driven phrase structure grammar. Studies in contemporary linguistics. Stanford: CSLI. Portner, Paul (2009). Modality. Oxford: Oxford University Press. Quirk, Randolph et al. (1985). A Comprehensive Grammar of the English Language. London: Longman. Raynaud, Franziska (1977). ‘Noch einmal Modalverben!’ In: Deutsche Sprache 1, pp. 1–30.

References |

573

Redder, Angelika (1984). Modalverben im Unterrichtsdiskurs. Pragmatik der Modalverben am Beispiel eines instituttionellen Diskurses. Vol. 54. Reihe germanistische Linguistik. Tübingen: Niemeyer. Reis, Marga (1995). ‘Über infinite Nominativkonstruktionen’. In: Festvorträge anläßlich des 60. Geburtstags von Inger Rosengren. Ed. by Olaf Önnerfors. Sprache und Pragmatik. Arbeitsberichte Sonderheft. Lund: University of Lund. Reis, Marga (2001). ‘Bilden Modalverben im Deutschen eine syntaktische Klasse?’ In: Modalität und Modalverben im Deutschen. Ed. by Marga Reis and Reimar Müller. Hamburg: Buske, pp. 239–262. Reis, Marga (2003). ‘On the form of German wh-infinitives’. In: Journal of Germanic Linguistics 15, pp. 155–201. Reis, Marga (2005a). ‘Wer ’brauchen’ ohne ’zu’ gebraucht: zu systemgerechten Verstössen im Gegenwartsdeutschen’. In: Cahier d’Etudes Germaniques 48.1, pp. 101–114. Reis, Marga (2005b). ‘Zur Grammatik der sogenannten ’Halbmodale’ ’drohen’/’versprechen’ + Infinitiv’. In: Deutsche Syntax: Empirie und Theorie, Symposium Göteborg 13-15 Mai 2004. Ed. by Franz-Josef d’Avis. Vol. 46. Acta Universitatis Gothoburgensis, pp. 125–145. Reis, Marga (2007). ‘Modals, so-called semi-modaly and grammaticalization in German’. In: Interdisciplinary Journal for Germanic Linguistics and Semiotic Analysis 12.1, pp. 1–58. Reis, Marga (2013). ‘„Weil-V2”-Sätze und (k)ein Ende? Anmerkungen zur Analyse von Antomo & Steinbach 2010’. In: Zeitschrift für Sprachwissenschaft 32.2, pp. 221–262. Reis, Marga and Angelika Wöllstein (2010). ‘Zur Grammmatik (vor allem) konditionaler V1Gefüge im Deutschen’. In: Zeitschrift für Sprachwissenschaft 29, pp. 111–179. Remberger, Eva-Maria (2010). ‘The evidential shift of WANT’. In: Evidence from Evidentials. Ed. by Tyler Peterson and Uli Sauerland. Vol. 25. UBC Working Papers in Linguistics. Vancouver: University of British Columbia, pp. 161–182. Richter, Frank and Manfred Sailer (2008). ‘Simple Trees with Complex Semantics: On Epistemic Modals and Strong Quantifiers’. In: Proceedings of the ESSLLI Workshop on ”What Syntax feeds Semantics?” Ed. by Maribel Romero. Hamburg. Riemsdijk, Henk van (2002). ‘The unbearable lightness of going: the projection parameter as a pure parameter governing the distribution of elliptic motion verbs in Germanic’. In: Journal of Comparative Germanic Linguistics 5, pp. 143–196. Rix, Helmut (2001). Lexikon der indogermanischen Verben. 2nd ed. Wiesbaden: Reichert. Roberts, Ian and Anna Roussou (1999). ‘A formal approach to ’grammaticalization’’. In: Linguistics 37.6, pp. 1011–1041. Roberts, Ian G. (2003). Synactic change: a minimalist approach to grammaticalization. Vol. 100. Cambridge studies in linguistics. Cambridge: University Press. Rooth, Mats (1992). ‘A theory of focus interpretation’. In: Natural Language Semantics 1, pp. 75–116. Rooth, Mats Edward (1985). ‘Association with focus’. PhD thesis. Amherst: University of Massachusetts. Ross, John Robert (1969). ‘Auxiliaries as main verbs’. In: Studies in Philosophical Linguistics 1, pp. 77–102. Rossmann, Bernhard (1908). Zum Gebrauch der Modi und Modalverba in Adverbialsätzen im Frühmittelenglischen. Kiel: Fiencke. Ruoff, Arno (1981). Häufigkeitswörterbuch gesprochener Sprache. Tübingen: Niemeyer. Sanders, Daniel (1908). Wörterbuch der Hauptschwierigkeiten in der deutschen Sprache. 38th ed. Berlin Schöneberg: Langenscheidt.

574 | References

Scaffidi-Abbate, August (1973). ‘’Brauchen’ mit folgendem Infinitiv’. In: Muttersprache 83, pp. 1–45. Schachter, Paul (1984). ‘A note on syntactic categories and coordination in GPSG’. In: Natural Language and Linguistic Theory 2, pp. 269–281. Schallert, Oliver (2014a). ‘IPP–constructions in Alemannic and Bavarian in comparison’. In: Bavarian. Ed. by Günther Grewendorf and Helmut Weiß. Vol. 220. Linguistik Aktuell/Linguistics Today. Amsterdam: Benjamins. Schallert, Oliver (2014b). Zur Syntax der Ersatzinfinitivkonstruktion: Typologie und Variation. Studien zur deutschen Grammatik 87. Tübingen: Narr. Schenner, Matthias (2009). ‘Semantic of evidentials’. In: Proceedings of ConSOLE XVI 2008. Ed. by Sylvia Blaho, Camelia Constantinescu and Bert Le Bruyn. Leiden: Universiteit Leiden, pp. 179–198. Scheutz, Hannes (1999). ‘‘weil’-Sätze im gesprochenen Deutsch’. In: Beiträge zur Oberdeutschen Dialektologie. Ed. by Claus Jürgen Hutterer and Gertrude Pauritsch. Göppingen: Kümmerle, pp. 85–112. Schmid, Tanja (2000). ‘Die Ersatzinfinitivkonstruktion im Deutschen’. In: Linguistische Berichte 183, pp. 325–351. Schoetensack, Heinrich August (1856). Grammatik der neuhochdeutschen Sprache. Erlangen: Ferdinand Enke. Scholz, Ulrike (1991). Wunschsätze im Deutschen – formale und funktionale Beschreibung. Vol. 265. Linguistische Arbeiten. Tübingen: Niemeyer. Scholze-Stubenrecht, Werner (2000). Duden – Die deutsche Rechtschreibung. 22nd ed. Mannheim: Duden Verlag. Schottel, Justus Georg (1663). Ausführliche Arbeit von der Teutschen HaubtSprache: worin enthalten Gemelter dieser Haubtsprache Uhrankunft, Uhraltertuhm, Reinlichkeit, Eigenschaft, Vermögen, Unvergleichlichkeit, Grundrichtigkeit. Braunschweig: Ziliger. Schwabe, Kerstin (1994). Syntax und Semantik situativer Ellipsen. Vol. 48. Studien zur deutschen Grammatik. Tübingen: Narr. Selkirk, Lisa O. (1984). Phonology and syntax. The relation between sound and structure. Vol. 10. Current studies in linguistics. Cambridge, Mass: MIT Press. Shopen, Tim (1973). In: Foundations of Language 10. Speyer, Augustin (2011). ‘Zur Integriertheit kausaler (Neben-)Sätze im Fruehneuhochdeutschen’. In: Sprachwissenschaft 36.1, pp. 53–84. Stalnaker, Robert (1978). ‘Assertion’. In: Syntax and semantics 9: Pragmatics. New York: Academic Press. Stechow, Arnim von (2003). ‘Schritte zur Satzsemantik.’ Stechow, Arnim von and Wolfgang Sternefeld (1988). Bausteine syntaktischen Wissens: ein Lehrbuch der generativen Grammatik. Opladen: Westdeutscher Verlag. Steinitz, Renate (1999). ‘Die Kopula ‘werden’ und die Situationstypen’. In: Zeitschrift für Sprachwissenschaft 18.1, pp. 121–151. Stephenson, Tamina (2007). ‘Judge Dependence, Epistemic Modals and Predicates of Personal Taste’. In: Linguistics and Philosophy 30, pp. 487–525. Sütterlin (1910). Die deutsche Sprache der Gegenwart. 3rd ed. Leipzig: Voigtländer. Sweet, Henry (1891). New English Grammar. Logical and Historical. Volume I. Introduction, phonology and accidence. Oxford: Clarendon Press. Sweetser, Eve (1990). From etymology to pragmatics. Metaphorical and cultural aspects of semantic structure. Vol. 54. Cambridge studies in linguistics. Cambridge: University Press.

References |

575

Sweetser, Eve and Barbara Dancygier (2005). Mental spaces in grammar: conditional constructions. Cambridge: Cambridge University Press. Szumlakowski, Irene (2010). ‘’Dann brauche ich nicht mehr zum Rathaus’ Eine weitere Modalverbeigenschaft des Verbs ’brauchen’’. In: Deutsche Sprache 38.1, pp. 70–80. Szumlakowski-Morodo, Irene (2006). ‘’Ich will weg, wohin musst du’ Modalverben mit Richtungsbestimmung und ohne Infinitiv’. In: Muttersprache 116, pp. 322–347. Takahaši, Terukazu (1984). ‘Über den subjektiven Gebrauch des Modalverbs ’brauchen’’. In: Sprachwissenschaft 9, pp. 21–22. Tancredi, Christopher (2007). A Multi-Model Modal Theory of I-Semantics. Part I: Modals. Ms. University of Tokyo. Thráinsson, Höskuldur and Sten Vikner (1995). ‘Modals and double modals in the Scandinavian Languages’. In: Working Papers in Scandinavian Syntax 55, pp. 51–88. Traugott, Elizabeth Closs (1985). ‘Conditional markers’. In: Iconicity in syntax. Ed. by John Haiman. Amsterdam: John Benjamins, pp. 289–307. Traugott, Elizabeth Closs (1989). ‘On the rise of epistemic meanings in English: an example of subjectification in semantic change.’ In: Language 65.1, pp. 31–55. Traugott, Elizabeth Closs (1997). ‘Subjectification and the development of epistemic meaning: The case of ‘promise’ and ‘threaten’.’ In: Modality in Germanic Languages. Ed. by Toril Swan and Olaf Jansen Westvik. Berlin: de Gruyter, pp. 185–210. Tregidgo, Philip (1982). ‘‘Must’ and ‘may’: demand and permission’. In: Lingua 56.1, pp. 75–92. Truckenbrodt, Hubert (2006). ‘On the semantic motivation of syntactic verb movement to C in German’. In: Theoretical Linguistics 32.3, pp. 257–306. Uhmann, Susanne (1998). ‘Verbstellung in ’weil’-Sätzen: Lexikalische Differenzierungen mit grammatischen Folgen’. In: Zeitschrift für Sprachwissenschaft 1.17, pp. 92–139. Vainikka, Anne and Martha Young-Scholten (1994). ‘Direct Access to X’-Theory: evidence from Turkish and Korean adults learning German’. In: Language Acquisition Studies in Generative Grammar. Ed. by Bonnie D. Schwartz and Teun Hoekstra. Benjamins, pp. 256–316. Vater, Heinz (1975). ‘’werden’ als Modalverb’. In: Aspekte der Modalität. Ed. by Heinz Vater and Joseph Calbert. Studien zur deutschen Grammatik 1. Narr, pp. 71–148. Vater, Heinz (2001). ‘‘Sollen’ und ‘wollen’ – zwei ungleiche Brüder’. In: Modalität und mehr. Ed. by Heinz Vater and Ole Letnes. Vol. 23. Fokus. Trier: Wissenschaftlicher Verlag, pp. 81– 100. Vater, Heinz (2004). ‘Zur Syntax und Semantik der Modalverben’. In: Modalität und Übersetzung. Ed. by Ole Letnes and Heinz Vater. Trier: Wissenschaftlicher Verlag, pp. 9–31. Vater, Heinz (2010). ‘’Möchten’ als Modalverb’. In: Modalität / Temporalität aus kontrastiver und typologischer Sicht. Ed. by Andrzej Kątny and Anna Socka. Vol. 30. Danziger Beiträge zur Germanistik. Peter Lang, pp. 99–112. Vernaleken, Theodor (1861). Deutsche Syntax. Vol. 1. Wien. Vogel, Petra (2005). ‘Neue Überlegungen zu den Fügungen des Typs ‘sie kamen gelaufen’’. In: Zeitschrift für Germanistische Linguistik 33, pp. 57–77. Watts, Richard J. (1984). ‘An analysis of epistemic possibility and probability’. In: English Studies 65.2, pp. 129–140. Wegener, Heide (1993). ‘weil - das hat schon seinen Grund. Zur Verbstellung in Kausalsätzen mit ’weil’ im Gegenwärtigen Deutsch’. In: Deutsche Sprache 21, pp. 289–305. Weinrich, Harald (1993). Textgrammatik der Deutschen Sprache. Mannheim: Dudenverlag.

576 | References

Welke, Klaus (1965). Untersuchungen zum System der Modalverben in der deutschen Sprache der Gegenwart. Vol. 10. Schriften zur Phonetik, Sprachwissenschaft und Kommunikationsforschung. Berlin: Akademie Verlag. Westmoreland, Robert R. (1998). ‘Information and intonation in natural language modality’. PhD thesis. Michigan: University of Michigan. Wilder, Chris (2008). ‘The PP-’with’-DP construction’. In: Elements of Slavic and Germanic grammars: a comparative view. Ed. by Jacek Witkos and Gisbert Fanselow. Frankfurt/Main: Peter Lang, pp. 235–253. Williams, Edwin (1983). ‘Against small clauses’. In: Linguistic Inquiry 14, pp. 287–308. Wilmanns, Wilhelm (1906). Deutsche Grammatik Gotisch, Alt-, Mittel- und Neuhochdeutsch. Vol. 3.1. Strassburg: Trübner. Winkler, Susanne (2006). ‘Ellipsis’. In: The Encyclopedia of Language and Linguistics. Ed. by Keith Brown. 2nd ed. Oxford: Elsevier. Wurmbrand, Susi (1999). ‘Modalverbs must be raising verbs’. In: Proceedings of the 18th West Coast Conference on Formal Linguistics. Ed. by Sonya Bird et al. Somerville, Mass: Cascadilla Press, pp. 599–612. Wurmbrand, Susi (2001). Infinitives. Restructuring and Clause Structure. Vol. 55. Studies in Generative Grammar. Berlin: Mouton de Gruyter. Wurzel, Wolfgang Ullrich (1984). Flexionsmorphologie und Natürlichkeit: ein Beitrag zur morphologischen Theoriebildung. Vol. 21. Studia grammatica. Berlin: Akademie Verlag. Wustmann, Gustav (1896). Allerhand Sprachdummheiten. 2nd ed. S. 334 anti FVG. Leipzig: Grunow. Wustmann, Gustav (1908). Allerhand Sprachdummheiten. 4th ed. Leipzig: Grunow. Zaefferer, Dietmar (1987). ‘Satztypen, Satzarten, Satzmodi – was Konditionale (auch) mit Interrogativen zu tun haben’. In: Satzmodus zwischen Grammatik und Pragmatik. Ed. by Jörg Meibauer. Tübingen: Niemeyer, pp. 259–285. Ziegeler, Debra (2006). ‘Omnitemporal ’will’’. In: Language Sciences 28, pp. 76–119. Zifonun, Gisela (1997). Grammatik der deutschen Sprache. Berlin: de Gruyter. Zimmermann, Malte (2004). ‘Zum Wohl: Diskurspartikel als Satztypmodifikatoren’. In: Linguistische Berichte 199, pp. 253–286.

Author index Abraham, Werner 1, 7, 14, 16, 17, 21, 32, 43, 48, 54, 77, 99, 108, 131, 210, 236, 255, 280, 323, 335, 338, 340, 342, 522, 523, 542, 546, 551 Adéwo.lé, Fé.mi O. 327, 328 Adelung, Johann Christoph 19, 43, 181, 186, 189, 217, 218, 259, 393 Aijmer, Karin 380, 395, 398, 406, 410, 415, 416, 426, 433, 441–443 Alban, Doctor 14 Aldenhoff, Jules 17, 18, 192, 195, 198 Allard, Francis X. 102, 155, 159, 165, 372, 400, 456 Ammann, Andreas 288, 328 Anand, Pranav 415 Antomo, Mailin 294, 296, 416, 546 Asbach-Schnitker, Brigitte 424, 438, 444 Askedal, John Ole 21, 22, 32, 57, 177, 184, 195, 206, 210, 211, 214, 215, 229–233, 235, 237–239, 242, 244, 245, 247, 248, 338, 342, 361, 373, 503, 504, 517, 538 Axel, Katrin 1, 33, 45, 47, 48, 77, 99, 165, 170, 348, 389, 396, 405, 541, 542 Bödiker, Johann 14, 19, 258, 273 Bühler, Karl 32, 336, 522 Barbiers, Sjef 48, 54–56, 58, 59, 61, 62, 77, 121, 280, 285, 286, 330, 345, 360 Bartsch, Renate 455 Basler, Otto 267, 268, 271 Bauer, Friedrich 8, 220, 226, 263, 267–269, 275 Bauer, Heinrich 262 Bech, Gunnar 8, 9, 22, 27, 30, 31, 59, 62, 95, 99, 100, 104, 108, 109, 115, 116, 118, 119, 123, 129, 131, 133, 137, 138, 142, 146, 147, 152, 154, 155, 165, 177, 184, 204, 238, 242, 269–272, 275, 286, 372, 393, 450, 456, 496, 523 Becker, Karl Ferdinand 17, 29, 30, 44, 76, 98, 99, 108, 118, 124, 129, 137, https://doi.org/10.1515/9783110411027-011

142, 147, 148, 155, 178, 184, 190, 260–265, 269, 270, 273, 276, 496 Behaghel, Otto 57, 267, 308 Beringer, Nicole 17 Bhatt, Rajesh 431 Bildhauer, Felix 532 Birkmann, Thomas 10, 11, 17, 23, 24, 33, 34, 43, 116, 193 Blühdorn, Hardarik 50, 134, 374 Blatz, Friedrich 264, 269, 273, 275 Bloch, Oscar 183 Bolkestein, Machtelt A. 325, 326, 542 Boye, Kasper 1, 70, 177, 178, 323, 340, 360, 455, 485 Brünner, Gisela 218, 272 Braune, Wilhelm 10, 11, 266, 273, 276 Brennan, Virginia 38, 45, 47, 48, 52, 54, 72, 77, 81, 208, 220, 224, 226, 228, 287, 288, 323, 329, 378, 385, 386, 394, 453, 472, 476, 523 Bres, Jacques 220 Brinkmann, Hennig 100, 176, 177 Buscha, Joachim 8, 9, 12, 26, 40, 104, 123, 172, 342 Butler, Jonny 323, 328, 378, 473, 538 Butt, Miriam 256 Bybee, Joan 1, 541 Calbert, Joseph P 100 Carlson, Gregory N. 38, 52, 81, 208, 288, 549 Carpenter, Bob 80, 118, 132, 206 Chomsky, Noam 233 Cinque, Guglielmo 6, 341, 380, 389, 394, 455, 492, 535, 537 Clément, Danièle 307, 455 Claius, Johannes 10, 19, 20, 257 Coates, Jennifer 64, 72, 287, 348, 355, 357, 372–375, 380, 472, 481 Cohen, Ariel 335, 467 Colomo, Katarina 109, 141, 235, 238, 239, 241, 242, 244, 248, 249, 286, 333, 341, 348, 504, 514, 526, 527, 535

578 | Author index

Comrie, Bernard 227 Condoravdi, Cleo 355 Coniglio, Marco 414, 434, 436, 438, 441, 444 Copley, Bridget 71, 75, 144, 145 Cornillie, Bert 239, 323, 326 Curme, George O. 8, 9, 11, 31, 108, 124, 137, 165, 265, 266, 269, 275, 287, 393, 394, 486

Dal, Ingerid 270 Dancygier, Barbara 289 De Haan, Ferdinand 327, 328 Declerck, Renaat 429, 432 Demirdache, Hamida 323, 345 Demske, Ulrike 23, 24 Denison, David 55, 130, 219, 542, 543 Depraetere, Ilse 386, 524 Diesing, Molly 549 Dietrich, Rainer 387 Diewald, Gabriele 1, 3, 8, 9, 23, 24, 27, 31, 33, 34, 37, 44, 46, 48, 54, 62, 99, 102, 119, 128, 146, 147, 152, 154, 155, 159, 165, 166, 170, 229, 230, 232, 233, 235, 244, 245, 247, 250, 253, 255, 275, 323, 348, 372, 373, 389, 400, 402, 403, 405, 406, 456, 464, 466–471, 475, 476, 523, 524, 527, 541, 543 Dik, Simon C. 332 Doherty, Monika 386, 485, 496 Doitchinov, Serge 73, 74, 542 Dougherty, Ray C. 97, 137 Dowty, David 111 Drosdowski, Günther 199, 242, 271, 272 Drubig, Hans Bernhard 72, 288, 328, 335, 361, 373, 378, 380, 389, 426, 444, 472, 481 Duden, Konrad 267, 268

Ebert, Robert Peter 13, 14, 16, 46, 167, 229, 233, 235 Ehlich, Konrand 33, 211 Ehrich, Veronika 100, 104, 109, 133, 141, 174, 503, 526

Eide, Kristin Melum 6, 63, 142, 178, 323, 335, 338, 341, 345, 360, 361, 368, 481, 485, 488, 492, 535 Eisenberg, Peter 8, 11, 19, 22, 95, 98, 147, 167, 168, 176, 220, 223, 272, 277, 299, 304, 306, 307, 314, 342, 343, 355, 429, 432–434, 466, 488, 501 Enç, Mürvet 32, 220, 224, 226, 228, 286, 401, 453 Engel, Ulrich 8, 22, 31, 32, 35, 176, 226, 250, 275, 364 Erb, Marie Christine 19, 31, 32, 34, 41, 44–48, 54, 56, 60, 76, 77, 95, 98, 99, 110, 111, 170, 223–228, 323, 335, 338, 339, 342, 343, 348–350, 360, 368, 371, 481, 524, 537–539 Eroms, Hans-Werner 294 Fabricius-Hansen, Cathrine 151, 227, 228, 292, 293, 429, 430, 432, 440, 466 Fagan, Sarah M. B. 342, 490, 504 Faller, Martina 109, 141, 526, 527, 529 Feldman, Fred 48, 54 Fillmore, Charles J. 31, 336, 522 Fintel, Kai von 31, 62, 354, 355, 378, 389, 408, 444–447, 452, 458, 473, 477, 524, 535 Fleischer, Jürg 16 Folsom, Marvin H. 176, 177, 184, 193, 198, 204, 206, 210, 211, 275 Fourquet, Jean 271, 323 Freywald, Ulrike 302 Fritz, Gerd 1, 2, 27, 33, 34, 43, 52, 55, 64, 67, 70, 93, 95, 98, 115, 121, 128, 131, 133, 142, 147, 154, 155, 160, 164, 165, 226, 228, 287, 372, 373, 393, 394, 456, 464, 466, 541, 543 Gamon, David 468 Gazdar, Gerald 97, 137 Geilfuß, Jochen 48, 77 Gelhaus, Hermann 199 Gergel, Remus 25, 48, 103, 104, 111 Geuder, Wilhelm 332 Gillies, Anthony S. 31, 62, 354, 355, 378, 408, 477, 524, 535

Author index |

Ginzburg, Jonathan 318 Girnth, Heiko 17, 190, 192, 193 Glas, Reinhold 131, 142, 432 Götzke, Alfred 199, 201, 202, 267 Grebe, Paul 193, 242, 270, 271 Grice, Paul 238 Gries, Stefan Th. 42 Griesbach, Heinz 8, 9, 21, 31, 270, 272, 275, 342 Grimm, Jacob 9, 11, 14, 15, 17, 18, 35, 178, 181, 182, 184, 186, 190, 195, 217, 218, 259, 260, 277 Grimm, Wilhelm 181, 182, 186, 217, 218 Grosz, Patrick 50, 488, 535 Gunlogson, Christine 319 Günthner, Susanne 298–302, 416 Hacquard, Valentine 280, 288, 323, 337, 345, 349, 354, 383, 394, 411, 415, 432, 452, 482, 484, 523 Haegeman, Liliane 289–294, 301, 302, 414, 428, 429, 432–434, 441, 443, 466 Haider, Hubert 37, 249 Hansen, Björn 327 Hartmann, Jutta 25, 48, 103, 104, 111 Heim, Irene 446 Heine, Bernd 212, 225, 349 Heinrich, Gertraud 8 Helbig, Agnes 310 Helbig, Gerhard 8, 9, 12, 26, 40, 104, 123, 172, 310, 342 Hengeveld, Kees 348, 373, 380, 426, 468 Hetland, Jorunn 20, 43, 54, 55, 58, 60, 142 Heyse, Johann Christian August 18, 38, 54, 260, 261 Hill, Virginia 326 Hinterhölzl, Roland 12, 13 Hinterwimmer, Stefan 99, 131, 138, 145 Hofmann, Thomas Ronald 72, 287, 378, 472 Höhle, Tilman N. 37, 48, 77, 99, 131, 369, 235, 369, 371 Holler, Anke 189 Homer, Vincent 349, 354, 378, 473 Hornstein, Norbert 46

579

Huitink, Janneke 126, 127, 163, 335, 374, 406, 410, 411, 415, 426, 446–449, 453–456, 467, 470, 471, 473, 475, 477 Iatridou, Sabine 378, 389, 444–447, 452, 458, 473 Jäger, Agnes 233 Jäger, Gerhard 331 Jäger, Siegfried 185, 198 Jackendoff, Ray 46, 48, 54, 77, 97, 137, 323, 380, 386, 389 Jacobs, Joachim 369, 370, 374, 532, 533 Jędrzejowski, Łukasz 344, 354 Jellinek, Max Hermann 257, 260 Jespersen, Otto 55, 267 Johnen, Thomas 27, 260 Kaeding, Friedrich Wilhelm 37 Kaiaty, Mohamed 151, 152 Kaita, Kousuke 219 Kasper, Walter 66, 90, 342, 345, 355 Kiaulehn, Walther 193 Kiefer, Ferenc 335, 368, 373, 380, 402, 469 Kirchner, Gustav 269 Kiss, Tibor 33, 37, 46, 48, 77, 99, 100, 128, 165, 170, 238 Kissine, Mikhail 224 Klein, Wolfgang 348 Klindt, Saskia 288, 328 Kluempers, John David 182, 202, 211, 218, 219 Kluge, Friedrich 179 Kolb, Hermann 176, 177, 179, 184, 190–192, 194, 195, 204, 217, 218, 271, 275 König, Ekkehard 307, 545 Kotin, Michail L. 124, 160, 327 Krämer, Sabine 3, 31, 223, 225–228, 335, 337, 405, 406, 415, 442, 524 Kramer, Matthias 189 Kratzer, Angelika 7, 27–29, 31, 34, 62, 113, 118, 119, 123, 124, 158, 224, 228, 272, 286, 289, 290, 313, 317, 323, 328, 329, 331, 333, 371, 377,

580 | Author index

384, 428, 433, 441, 456, 471, 495, 523, 524, 531, 537, 542, 549 Krause, Maxi 542, 543 Krifka, Manfred 370 Kurrelmeyer, Wilhelm 13, 16, 268 Kürschner, Wilfried 204, 210, 211 Labeau, Emanuelle 220 Lang, Ewald 424 Langer, Nils 21, 198 Lasersohn, Peter 386, 402, 520, 524 Lassiter, Daniel 28, 29, 289, 312, 323, 531, 532 Lechner, Winfried 455, 456 Leech, Geoffrey N. 158, 287, 372, 380, 386, 430, 444 Lehmann, Christian 253, 255, 541 Leiss, Elisabeth 7, 280, 542, 546, 551 Lenz, Barbara 135, 177, 204, 364 Letnes, Ole 487 Levinson, Stephen 31, 336, 522 Levinson, Stephen C. 118, 124, 125, 160, 161 Lewis, David 495, 537, 542, 549 Leys, Odo 305–307 Lightfoot, David 12, 550 Lima, José Pinto de 230, 238, 239 López, Luis 39, 361 Lötscher, Andreas 63, 64, 66, 67, 85, 123, 148, 165, 450 Lyons, John 6, 26, 38, 62, 323, 334, 368, 369, 371–373, 375, 378, 380, 385, 389, 398, 402, 406, 409, 412, 415, 425, 427, 429, 432, 446, 457, 459, 461–473, 479, 481, 533 Maché, Jakob 14, 16, 17, 21, 210, 236, 286, 297, 322, 329, 539, 546 Maienborn, Claudia 7, 42, 220, 286, 289, 313, 317, 328, 331, 333, 549 Maitz, Péter 177, 193 Martin, Fabienne 31, 62, 72, 349, 525 McDowell, Joyce 380, 389, 450, 532 Meid, Wolfgang 9 Menedéz-Benito, Paula 457, 458, 546 Merchant, Jason 39 Milan, Carlo 323

Mitka, Walther 199, 201, 202, 267 Mortelmans, Tanja 1, 67, 68, 70, 88, 93, 177, 178, 340, 355, 360, 455, 485 Moscati, Vincenzo 378, 473 Moulton, Keir 455 Müller, Reimar 2, 543 Müller, Stefan 37, 45, 220, 221, 223, 242, 243 Murray, James A.H. 193, 218, 219 Neugeborn, Wolfgang 237 Newmeyer, Frederick J. 46, 48 Nissenbaum, Jon 488, 535 Nuyts, Jan 323, 380, 411, 412, 416, 424–426, 447, 454, 467–470, 475, 477, 478 Öhlschläger, Günther 1, 25, 31–33, 38, 39, 44, 46, 48, 52, 57, 72, 73, 77, 95, 96, 98–100, 102, 108–110, 116, 118, 119, 121, 123, 128, 131–134, 141, 142, 147, 152, 155, 158, 165, 166, 169, 170, 173, 174, 176, 177, 211, 220, 250, 260, 275, 323, 368, 369, 371–374, 376, 377, 384, 385, 398–402, 405, 411, 416, 426, 431, 453, 456, 459, 460, 467–469, 471–473, 475, 483, 485, 501, 503, 527 Ölinger, Albert 19, 257 Ørsnes, Bjarne 54, 56, 60, 192 Pafel, Jürgen 229, 230, 232, 235, 237 Pagliuca, William 1, 541 Palmer, Frank Robert 26, 45, 48, 62, 77, 109, 142, 285, 323 Pancheva, Roumyana 431 Papafragou, Anna 31, 124, 160, 335, 337, 378, 380, 398, 402, 406, 409–415, 419, 426, 432, 460, 467, 470, 473, 477, 483 Paul, Hermann 43, 54, 56, 57, 177, 179, 181, 184, 203, 217, 266, 541 Perkins, Michael R. 131, 468, 469 Perkins, Revere Dale 1, 541 Perlmutter, David 45 Pfeffer, Alan J. 184, 185, 202, 204

Author index |

Pietandrea, Paola 323 Pillot, Jean 257 Pittner, Karin 306 Pokorny, Julius 10 Pollard, Carl 45, 46, 97, 137, 220 Portner, Paul 26, 54 Quirk, Randolph 299 Raynaud, Franziska 95, 98, 99, 123, 131, 168, 169, 211, 212, 285, 286, 339 Redder, Angelika 9, 260, 272 Reed, Susan 429, 432 Rehbein, Jochen 33, 211 Reiffenstein, Ingo 10, 11 Reis, Marga 5, 31, 32, 35, 46–48, 51, 54, 59, 77, 98–100, 108, 110, 128, 141, 165, 166, 169, 170, 177, 184, 214, 215, 217, 220, 238, 239, 241–245, 248–250, 253, 275, 277, 279, 294, 296, 305, 308, 316, 323, 337, 339, 340, 343, 344, 348, 362, 431, 432, 485, 489, 495, 496, 501, 504, 508, 513, 514, 517, 536 Remberger, Eva-Maria 100 Richter, Frank 455 Riemsdijk, Henk van 55–57, 361 Rix, Helmut 10 Roberts, Ian G. 323, 538, 550 Rooth, Mats Edward 370 Ross, John Robert 48, 77, 116, 323, 361, 363 Rossmann, Bernhard 269, 275 Roussou, Anna 550 Ruoff, Arno 35, 36, 147, 212 Sag, Ivan A. 45, 46, 97, 137, 220 Sailer, Manfred 455 Sanders, Daniel 14, 18, 19, 178, 184, 187, 195, 197, 198, 202, 266 Scaffidi-Abbate, August 177–179, 183, 184, 189–194, 202, 203, 217, 218, 275 Schachter, Paul 97, 137 Schallert, Oliver 16 Schenner, Matthias 109, 415, 499 Scheutz, Hannes 296, 319

581

Schmid, Tanja 17, 18, 194, 198 Schoetensack, Heinrich August 18, 59, 102, 108, 109, 124, 137, 155, 190, 262, 263, 269 Scholz, Ulrike 142, 396–398, 492 Scholze-Stubenrecht, Werner 193 Schottel, Justus Georg 257, 258 Schulz, Dora 8, 9, 21, 31, 270, 272, 275, 342 Schwabe, Kerstin 39 Selkirk, Lisa O. 369, 374 Shopen, Tim 57 Smirnova, Elena 244, 245, 247 Speyer, Augustin 296 Stalnaker, Robert 533 Stechow, Arnim von 46–48, 50, 54, 79, 100, 116 Stefanowitsch, Anatol 42 Steinbach, Markus 294, 296, 416, 546 Steinitz, Renate 220 Stephenson, Tamina 68, 401, 461, 520, 524 Sternefeld, Wolfgang 47, 48, 54, 100 Sütterlin 266 Sweet, Henry 18, 193, 265 Sweetser, Eve 1, 72, 228, 229, 287, 289, 290, 294, 296, 322, 323, 373, 378, 384, 416, 427, 428, 472, 541 Szumlakowski, Irene 176, 177, 210 Szumlakowski-Morodo, Irene 54, 55, 60, 98, 121, 170, 171 Sæbø, Kjell Johan 292, 293, 429, 430, 432, 440, 466

Takahaši, Terukazu 32, 177, 184, 211, 215 Tancredi, Christopher 446–448, 454–456, 458, 467–469, 471, 473 Thümmel, Wolfgang 455 Thráinsson, Höskuldur 110, 323, 338, 368 Traugott, Elizabeth Closs 1, 239, 431, 541 Tregidgo, Philip 322 Tronka, Krisztián 177, 193 Truckenbrodt, Hubert 109, 138, 319, 485

582 | Author index

Uhmann, Susanne 294–296, 298, 302, 416, 419 Uribe-Etxebarria, Myriam 323, 345 Vainikka, Anne 204 van der Auwera 1, 70, 177, 178, 288, 307, 328, 340, 360, 455, 485, 545 Vater, Heinz 20, 32, 34, 43, 54, 55, 58, 60, 98, 124, 128, 142, 166, 169, 173, 175, 211, 220, 223, 224, 226, 228, 275, 360, 368, 401 Verhulst, An 386, 524 Vernaleken, Theodor 8, 31, 263, 265, 269, 275, 489 Vikner, Sten 110, 323, 338, 368 Vogel, Petra 59 Wartburg, Walther von 183 Watts, Richard J. 380, 402, 468, 469, 471, 472 Wegener, Heide 289, 294–296, 301, 416, 417 Weinrich, Harald 123, 152, 155, 176 Welke, Klaus 19, 21, 22, 26, 27, 33, 37, 40, 47, 48, 77, 82, 98–100, 102, 116, 121, 123, 131, 138, 152, 154, 155, 159, 160, 166, 169, 176, 210, 218, 226, 270, 271, 275, 286, 372, 450, 456, 486, 487 Wellwood, Alexis 288, 337, 383, 394, 411, 415, 432, 452, 482, 484

Westmoreland, Robert R. 31, 62, 379, 474, 523, 525 Wilder, Chris 55, 56 Williams, Edwin 455 Wilmanns, Wilhelm 266 Winkler, Susanne 39, 361 Wöllstein, Angelika 305, 308, 316, 431, 501 Wurmbrand, Susi 25, 32–34, 41, 45, 47, 48, 50, 77, 79, 99, 100, 121, 128, 132, 165, 170, 171, 229, 237–239, 242, 247, 248, 275, 323, 335, 338, 342, 481, 503, 504, 517, 537–539 Wurzel, Wolfgang Ullrich 17, 24, 34, 193 Wustmann, Gustav 187, 195, 197, 198, 218 Young-Scholten, Martha 204 Zaefferer, Dietmar 431 Ziegeler, Debra 31, 62, 150, 228, 286, 474, 524 Zifonun, Gisela 8, 20, 22, 35, 41, 42, 54, 70, 95, 123, 133, 160, 175, 211, 224, 228, 272, 286, 298, 338, 342, 348, 357, 359 Zimmermann, Malte 62, 289, 310, 335, 337, 387, 388, 392, 395, 398, 404, 409, 423, 433, 444, 484, 520 Zoch, Irene 8

Subject index accommodation (rule of) 495, 537, 542 adjective 26, 28, 42, 71, 191, 220, 230, 231, 424, 425, 469, 470, 532 adverb 26, 50, 88, 102, 104, 107, 118, 143, 160, 164, 169, 171, 181, 224, 228, 246, 280, 289, 298, 309–316, 321, 333, 373, 380, 389, 398, 402, 403, 414, 416, 424, 425, 429, 433, 450, 455, 457, 458, 468–470, 472, 475, 480, 532 adverbial 5, 6, 22, 63, 68, 151, 181, 280, 289, 294, 295, 298, 300, 303–312, 316–322, 333, 335, 337, 341, 371, 386–389, 402, 418, 421, 422, 425, 429, 430, 433–438, 446, 457, 469, 474, 479, 481–488, 502, 518, 532, 534–536, 539, 553 adverbial clause 151, 289, 294, 295, 300, 303–306, 309, 316, 337, 371, 418, 421, 422, 425, 429, 430, 433–436, 438, 474, 479, 482, 484 auxiliary 6, 12, 13, 16–23, 26–30, 34–37, 42, 55, 57, 70, 71, 89, 176–178, 181, 184, 190, 198, 212, 218–224, 237, 242, 257–277, 290, 323–326, 338–343, 345, 348, 350, 354, 355, 372, 378, 380, 389, 395, 400, 401, 419, 433, 442, 445, 446, 453, 461, 481, 486, 489, 502, 503, 508, 517, 518, 535–539, 547, 553 – future 6, 21, 220, 223, 224, 339, 341, 400, 401, 453 – modal 13, 21, 23, 26–29, 34, 37, 71, 177, 218, 219, 223, 224, 257, 262–269, 273–277, 290, 342, 348, 372, 378, 380, 389, 395, 419, 433, 442, 445, 446 – passive 42, 55, 181, 220–223, 257 – past 70, 89

causal clause 280, 294–298, 322, 333, 416–419, 430, 482 – epistemic 297, 298, 416 https://doi.org/10.1515/9783110411027-012

– event-related 294–298, 322, 416–419, 430, 434, 482, 503 circumstantial (modal[ity]) 1, 3, 5–7, 28, 32, 37, 48, 49, 52–58, 62, 63, 77, 84, 109, 113, 116, 119–122, 131–138, 142, 143, 146, 147, 150, 154–156, 159, 177, 178, 197, 206, 214, 220, 224, 226, 238, 250–252, 255, 256, 280, 284–290, 293, 299, 315, 321–335, 340–342, 346, 355, 361, 363, 368, 369, 372, 379, 380, 396, 397, 430, 457, 477, 479, 481, 483, 485, 492, 502, 518, 523, 535–542, 546–554 CoDeC (condition on deictic centres) 32, 33, 35, 62, 72, 84, 109, 123, 141, 144, 146, 158, 212, 238, 250, 252, 256, 288, 341, 379, 385, 450, 518, 524–526, 530, 534, 536 concessive epistemic modality 102, 112, 113, 141, 157–159, 165, 353, 545 conditional 26, 66, 67, 69, 89, 90, 94, 144, 145, 152, 180, 265, 280, 289–293, 299, 321, 326, 333–337, 343, 357, 386, 396, 398, 419, 425–434, 440, 441, 463, 466, 474, 475, 481–486, 500–503, 511–513, 517, 518, 530, 534–536, 545 conditional clause – echoic 292, 293 – epistemic 66, 67, 290–293, 428, 501 – event-related 290–293, 335, 419, 425, 428–433, 441, 466, 481, 482, 486, 500–503, 511–513, 517, 518, 535, 536 – irrelevance 427, 433 context shift 279, 320, 336, 340–343, 348–353, 384, 386–389, 395, 402, 405, 415, 419, 425, 436, 476, 486, 492, 495, 499–501, 514, 518, 520 control (predicate/verb) 25, 33, 37, 38, 45–49, 54, 60, 61, 74–77, 99, 100, 110–113, 116, 133, 147, 170, 174, 184, 206, 207, 239, 241, 253, 254,

584 | Subject index

304, 307, 332, 333, 368, 489, 507, 537, 538, 541, 550 counterfactual 63, 64–68, 70, 71, 75, 88–91, 94, 144, 145, 148, 150, 165, 235, 238, 342, 343, 345, 355, 357 de dicto, 45, 46, 50, 76, 116, 132, 206, 445, 452, 455 de re, 46, 50, 144, 386, 445, 446, 452, 455 deictic centre 6, 31, 32, 62, 68, 72, 84, 91, 143, 158, 250, 279, 320, 321, 336, 337, 340–345, 348, 349–353, 363, 384–386, 388, 389, 394, 395, 401, 402, 405, 408–412, 414–419, 422–425, 430–433, 435–441, 444, 469, 476–478, 480–486, 491–496, 499–502, 511–513, 518, 521–526, 530, 533–539, 540, 553, 554 directional phrase 5, 37, 38, 54–63, 75–77, 83, 100, 107, 121, 136, 137, 154, 155, 170, 171, 177, 178, 210, 220, 230, 278, 335, 360, 361, 480, 483, 518, 534, 553 – verbless 5, 38, 55–63, 75–77, 83, 100, 107, 121, 136, 137, 154, 155, 170, 171, 177, 178, 210, 278, 360, 361, 480, 483, 518, 534, 537, 553 English V 1, 11, 12, 17, 28, 29, 39, 42, 52, 55, 64, 71, 72, 81, 130, 131, 140, 144, 158, 165, 177, 179, 193, 218–220, 224–228, 231, 239, 265–267, 276, 287–291, 294, 296, 301, 316, 323, 326, 327, 330, 340, 348, 354, 355, 361, 369, 372, 374, 378–380, 383, 389, 394, 395, 398, 402, 406, 410, 411, 415, 419, 426, 433, 434, 437, 441–448, 453, 456–458, 464, 468, 469, 482, 484, 508, 538, 542, 543, 547 epistemic 1–7, 22, 25, 28–40, 45, 48, 50, 55, 59, 62–75, 84–95, 102, 108–113, 119–126, 128–131, 138, 141–147, 155–167, 175–179, 184, 211–220, 223–226, 228–231, 235–239, 245–256, 261, 265,

271–280, 284–299, 302, 308–363, 368–389, 392–496, 499–504, 511–526, 530–554 epistemic modal – objective 62, 345, 369, 371–373, 385–389, 411, 423, 426, 429, 432, 463, 464, 467, 471, 474, 480, 553 – subjective 334, 368, 369, 371, 372, 374, 375, 411, 426, 461, 463, 473 event 5, 7, 9, 10, 14, 22, 26, 37, 45, 48, 54, 62, 75–77, 83, 84, 99–101, 107, 108, 116, 121, 122, 130, 136, 138, 142, 152–156, 159, 170–173, 184, 210, 212, 223, 226, 228, 241–253, 272, 280, 284–318, 321–338, 341, 345, 349, 351–355, 358, 362, 368, 401, 405, 416, 417–436, 440, 441, 466, 470, 477, 481–486, 492–495, 500–505, 511–518, 522, 531, 535, 536, 539, 541, 545, 546, 548–554 – argument 7, 286, 290, 294, 298, 306, 310, 313, 315, 317, 321, 327–333, 483, 549 – modifier 5, 7, 280, 286–288, 293, 304, 305, 317, 321, 324, 327–333, 368, 548–550, 553, 554 evidential 6, 27, 62, 126, 228, 229, 377, 386, 455, 459, 467, 470, 485, 503 focus 125, 134, 169, 225, 228, 235, 319, 320, 337, 362, 369–379, 383, 409, 450, 452, 531, 532 – contrastive 134, 169, 320, 362, 372–375, 377, 383 – verum 125, 134, 225, 228, 235, 337, 370, 371, 379 French 130, 140, 182, 183, 216–219, 220, 257, 259, 290, 317, 322–324, 327, 345, 354, 378 grammaticalisation 7, 34, 173, 184, 218, 247, 253, 255, 276, 464, 541, 548, 550, 551, 554 Hierarchy of Salience 533, 534, 540, 585 individual level predicates (ILP) 22, 26, 73, 317, 330–332, 585

Subject index |

infinitive 5, 6, 8, 10–26, 33–48, 54–63, 70, 75–77, 80, 83, 84, 89, 95–103, 107–116, 130, 131, 135, 137, 142, 147, 151–155, 166–206, 209–214, 217, 218, 220–230, 233–265, 270–273, 276–278, 286, 306–309, 316, 321, 322, 326, 327, 335–341, 349, 350, 354, 364, 368, 370, 389, 390, 396, 427, 475, 480–483, 486–489, 502, 507, 516, 518, 534–539, 541, 553 infinitive complement 5, 13–25, 34, 35, 38–44, 56–60, 63, 80, 96–98, 110–116, 130, 135, 137, 151–155, 168, 170, 177, 178, 180, 184, 185, 189–204, 210, 211, 214, 217–223, 226, 229, 230, 233, 235–241, 245–255, 261, 262, 265, 270–273, 277, 286, 326, 338–341, 349, 350, 354, 364, 368, 370, 389, 480, 483, 487, 488, 507, 516, 518, 534, 536, 537, 541, 553 infinitivus pro participio (IPP) 8, 9, 12–19, 29, 33, 35, 56, 176–180, 184, 190–200, 217, 218, 223, 236, 258, 262, 265–268, 271, 274, 277, 278, 342, 345, 346, 489, 508, 547 Kimean State Predicates 7, 546, 549 locative adverbial 280, 333 manner adverb 280, 333 modal base 28, 30, 118, 145, 146, 158, 272, 345, 371, 375, 380, 384, 523 modal force 28, 33, 77, 113, 115, 118, 123–128, 160, 164, 219, 224, 228, 272, 370, 371, 374, 375, 377–380, 384, 388, 452, 456, 457, 460, 461, 483, 523 modal particle 50, 362, 398, 441, 444 modal source 78, 95, 99, 121, 129, 131, 270, 272, 340, 386, 430, 431, 522, 523 modal verb 1–9, 11–48, 54–58, 61–68, 71–77, 81–88, 91–95, 99, 100, 103, 106–110, 113, 116, 118–151, 154,

585

155, 158–167, 172–179, 184, 190, 193, 194, 198, 203, 206, 212–220, 224–229, 235–239, 245–280, 285–294, 298, 299, 305, 308, 310, 313–317, 320–342, 345–355, 357–364, 368–380, 384–389, 392–400, 402, 405–426, 430–496, 499–504, 511–530, 533–543, 546–554 – traditional 4, 9, 12, 13, 16–27, 31–35, 38, 42–45, 48, 57, 61, 118, 130, 148, 164, 176, 177, 184, 190, 193, 194, 198, 203, 206, 220, 226, 228, 248, 250, 256, 257, 259–268, 277, 279, 338, 342, 503, 541, 552 motion verb 57, 59, 61

necessity 4, 24, 26–31, 62, 75–86, 91–95, 99, 100, 104, 106, 124, 128, 130, 135, 136, 160, 164, 172, 177, 178, 215, 219, 220, 224–228, 251, 261–264, 270, 273, 278, 323–327, 336, 341, 358, 369, 370, 377–379, 383, 386, 388, 393, 394, 400, 412, 452, 453, 456, 457, 462, 464, 466, 471, 474, 478, 480, 482, 496, 499, 500, 511, 513, 523, 524, 539, 542 negation 15, 50, 65–72, 75, 81, 82, 85, 86, 94, 102, 115, 125, 129, 133–136, 149–154, 174, 204, 206, 208, 212–215, 218, 219, 228, 265, 288, 335–338, 349, 350, 359, 364, 371–380, 383, 386, 388, 402, 448, 454, 462, 472–475, 479, 482–484, 503 Norwegian 6, 55, 58, 63, 120, 142, 178, 323, 338, 345, 361, 368, 481, 485, 488 noun phrase (NP) 38–43, 53, 76, 77, 79–83, 95–97, 100, 103, 110–117, 121, 126, 129, 147, 162, 164, 168, 174, 178–183, 191, 206, 217–223, 229, 231–242, 254, 285, 288, 329, 330, 360, 364, 401, 444, 447, 449, 452, 454–458, 489, 513, 516, 517, 538, 541

586 | Subject index

ordering source 28, 100, 118, 158, 523 polyfunctional 251–253, 256, 542 possibility 4, 7, 26–30, 33, 38, 40, 45, 46, 49–54, 60–64, 67, 70–72, 75, 81–86, 94, 95, 104, 113, 116–119, 124–126, 128, 147, 154, 158–165, 172, 175, 176, 219, 224, 250, 251, 261, 263, 264, 267, 270, 273, 278, 287, 288, 323, 326–328, 336, 338, 355, 369, 370, 374, 375–379, 383, 385, 388, 393, 394, 397, 400, 412, 447, 450, 452, 453, 456–458, 461–464, 466, 469, 471–478, 480, 482, 546, 547, 550 prepositional phrase (PP) 22, 55, 180, 181, 239 preterite present 8–16, 19, 20, 23, 24, 33, 34, 124, 184, 189, 190–194, 203, 219, 251, 260–268, 271, 276, 277, 395, 552 probability 29, 30, 93, 113, 123–128, 160, 220, 224, 228, 261, 456, 531 prohibition on vacuous quantification 358, 359, 462 quantificational modal 54, 83 question 31, 39, 72, 97, 110, 129, 130, 141, 189, 204, 228, 248, 291, 296, 303, 311, 314, 316–320, 334–337, 380–389, 392–396, 412, 414, 431, 462, 463, 472–476, 479, 480–486, 496–502, 518–522, 526, 533, 534 – polarity 311, 380, 383, 384, 387–389, 394, 395, 472, 482, 496 – wh, 39, 130, 204, 311, 319, 336, 337, 385, 389, 392, 394, 395, 482, 496, 497, 533 raising (predicate/verb) 15–18, 21–25, 37, 38, 45, 48, 49, 52, 54, 60, 61, 75–81, 100–104, 106, 107, 110–113, 116, 118, 122, 132, 135, 136, 141, 147, 154, 155, 158, 171–173, 177, 178, 183, 184, 191, 203, 206, 207, 210, 211, 217, 220, 222, 229, 230, 235–239, 242–249, 252–255, 258, 276–278, 341, 368,

371, 424, 454–456, 459, 503, 504, 507, 508, 511–514, 517, 537–539, 541 reportative modal(ity) 35, 140, 142, 174, 251, 284, 499, 528, 529, 533, 540 rule of accommodation 495, 537, 542 semantic role 46, 80, 99, 111, 118, 133, 368 Spanish 216, 239, 323, 327, 345, 348, 373, 426, 504, 505 stage-level predicate (SLP) 317, 331, 332 subjunctive 11, 21, 33, 63–71, 75, 85, 86–95, 113, 119, 122, 128, 129, 140–150, 164–166, 176, 180, 189–193, 218, 219, 251, 253, 260, 262, 266, 270, 273, 275, 287, 326, 339, 342, 343, 345, 355, 357, 378, 432, 448, 449, 483, 489, 503, 504, 508, 525, 543, 546, 547, 550 tense 10–13, 19, 26, 35, 56, 140, 176, 180, 190, 193, 212, 220, 223, 224, 248, 257–264, 277, 317, 321, 324, 342, 345, 348, 349, 353–355, 370, 481, 486, 489–492, 495, 502–504, 507, 508, 514–518, 547, 553 – future 220, 223, 257, 263, 264 – past 11, 12, 180, 190, 248, 277, 317, 321, 348, 349, 353, 355, 486, 495, 502, 503, 514–518 Time of Evaluation 238, 348–351, 354, 419, 435–441, 492 Time of Situation 348 Time of Utterance 150, 226, 238, 286, 294, 311, 312, 321, 327, 348–351, 354, 419, 434, 436, 492, 515, 516 Topic Time 302, 311, 348, 349, 351, 353, 419, 429, 433–441, 493 traditional modal verb 4, 9, 12, 13, 16, 17, 19–27, 31–35, 38, 42–45, 48, 57, 61, 118, 130, 148, 164, 176, 177, 184, 190, 193, 194, 198, 203, 206, 220, 226, 228, 248, 250, 256–268, 277, 279, 338, 342, 503, 541, 552 verb cluster 199, 202